Inflation: Mechanics Welfare: Its and Costs

ARTHUR M.
OKUN
Brookings Institution
Inflation: Its Mechanics

and Welfare Costs
ECONOMISTSCONTINUEto be baffled by the mystery of inflationary re-

cession(or stagflation)that has been experiencedby industrializedcoun-
triesduringthe 1970s.Whatcausesinflationto churnon and on even after
excessdemanddisappears?This whodunitis the latestin a long seriesthat
reflectsthe inadequacyof currenttheory in explainingthe mechanicsof
how inflationproceedsonce it has begun.In this paper,I shall sketchan
interpretationof the persistenceof inflationbased on an interactionof
"customer"or "career"markets,in whichpricesand wagesdo not equate
supplyand demand,with "auction"markets,in whichpricesdo reliably
clearmarkets.
ChangingStyles in InflationTheory
The suggestedinterpretationdevelopedin this paper builds on past

theoreticalliterature,whichI shall brieflyreview as a foundationfor the
subsequentanalysis.
VINTAGE KEYNESIANISM
VintageKeynesianismof the WorldWarII periodoffereda comparative-

static view of price-levelequilibrium,given an "inflationarygap" and a
Note: I am indebted to Nancy Delaney for her assistancein the research,and to
Edward Gramlich and several participantsin the Brookings panel for constructive
criticism.
351
352 Brookings Papers on Economic Activity, 2:1975
dynamicview of the approachto that equilibrium.'Overa wide rangeof

outputfromverydepressedlevelsup to full employment,variationsin ag-
gregatedemandchangedoutputwithpricesandwagesessentiallyconstant.
On the otherhand,in the inflationaryzone above full employment,incre-
mentalaggregatedemandmerelybid up the pricelevel,leavingproduction
unchanged.Thus,the aggregatesupplycurverelatingpricesandoutputhad
a rightangle.And the task of optimumpolicy was to make aggregatede-
mandintersectsupplyat the vertexof the rightangle-an assignmentthat
couldbe difficultbut thatinvolvedno agonizingtradeoffor crueldilemma.
Givenan inflationarygap,equilibriummightbe restoredat a new higher
price level becauseof a nonaccommodatingmonetarypolicy, or the re-
ducedrealvalueof cashand governmentbonds(the Pigoueffect),or auto-
maticstabilizersin the fiscalsystem.If, however,theseequilibratingforces
werenonexistentor very weak,hyperinflationmightresult.
The dynamicdeterminacyof the pricelevelwas attributedto one or both
of two'basiclags-(1) spendingbehindincomeand(2) wagesbehindprices.
In the former,consumersand businessmenformulateddollarbudgetsfor
theperiodahead,andstuckto them;thus,a sufficientjump in the pricelevel
would curb the real volume of spendingto the level of full-employment
output.2The rate of inflationthen dependedon the size of the inflationary
gap and the length of the budgetinglag. In the latter,this period'swage
level was set equalto last period'smarginalrevenueproductof labor;real
wagesthus wouldbe squeezedby inflationand so wouldrealconsumerex-
penditures(even if they did not lag behindincome).
Muchlater,RobertClowerinterpretedtheselags as imperfectionsin the
processof marketrecontracting.In the case of the budgetinglag, develop-
mentsin today'sfactormarketsdo not get into today'spriceson product
markets.In the case of the wage lag, today's developmentsin product
marketscannotget into today'spriceson factormarkets.3 Theselags pro-
1. John MaynardKeynes,How to Pay for the War(Harcourt,Brace, 1940).See also,
for example,CarlShoup, Milton Friedman,and Ruth P. Mack, Taxingto Prevent Infla-
tion (ColumbiaUniversityPress, 1943); Ralph Turvey, "Some Aspects of the Theory
of Inflation in a Closed Economy," Econonmic Journal,vol. 61 (September1951), pp.
531-43; and the referencesin note 4 below.
2. That would not happen, however,if the privatedecisionmakersformulatedtheir
spendingplans as shoppinglists ratherthan dollarbudgets.
3. Robert Clower,"The KeynesianCounterrevolution:A TheoreticalAppraisal,"in
F. H. Hahn and F. P. R. Brechling,eds., The Theoryof InterestRates (St. Martin's,
1965),pp. 112-13, 121-24.The samekind of mechanismunderliesthe model in RobertJ.
Barro and Herschel I. Grossman, "A General DisequilibriumModel of Income and
Employment,"Americani EconomicReview,vol. 61 (March 1971),pp. 82-93.
ArthurM. Okun 353
duce interestingdifference-equation models of the inflationaryprocess.4
Yet theyhavenot beena fruitfulbasisfor quantification. Thelengthsof the
lags wereneverestimatedwith any precision;and no link was established
betweentherateof inflationandthe size of the inflationarygap.Indeed,the
veryexistenceof a budgetinglag was neverempiricallydemonstrated.And
whilewagesand pricesdemonstrablychasedeach other,the evidencedid
not establishthat real wages were depressedby inflationas the vintage
Keynesianwage lag insisted.5
Gradually,in responseto postwarexperience,economistsroundedthe
rightangleof the aggregatesupplycurve,convertedfull employmentinto
a zone, and developedthe conceptof a tradeoff.But the earlyfiftiesgave
themno reasonto puzzleoverthe momentumof risingpricesbecauseinfla-
tion screechedto a halt in 1951-a developmentthat still standsout in
retrospectas an intriguingfortuitousmystery.The eventsof 1956-57posed
another,unpleasant,mystery:whyinflationbeganso soon at aggregateun-
employmentand operatingratesthat apparentlydid not involvegenerally
excessivedemand.CharlesSchultzeanalyzedthatpuzzlewitha refinedver-
sion of the vintageKeynesianinflationmodel,stressingsectoralimbalances
in demandand asymmetricresponsesto excesssupplyand excessdemand.6
PHILLIPS APPROACH
The Phillips approachdepartedfundamentallyfrom vintage Keynes-

ianismby relatinga givenutilizationrateto a givenrate of inflationrather
than an equilibriumlevel of prices,thus positinga continuoustradeoffbe-
tweeninflationand unemployment.As JamesTobin put it, "ThePhillips
curvehas beenan empiricalfindingin searchof a theory. . ."ITobin'sown
4. For example, see Tjalling Koopmans, "The Dynamics of Inflation,"Review of
EconzomicStatistics,vol. 24 (May 1942),pp. 53-65; A. Smithies,"TheBehaviorof Money
National IncomeunderInflationaryConditions,"QuarterlyJournialof Economics,vol. 57
(November 1942), pp. 113-28; and Ralph Turvey, "Period Analysis and Inflation,"
Economica,n.s., vol. 16 (August 1949), pp. 218-27.
5. The frustrationsof one young economistwith these hypotheses(and his unproduc-
tive negativeconclusions)are evidentin ArthuLrM. Okun,"TheEffectsof Open Inflation
on AggregateConsumerDemand" (Ph.D. dissertation,ColumbiaUniversity, 1956).
in the United States, Study Paper No. l of
6. Charles L. Schultze, Recent In2flation2
materialspreparedin connection with the Study of Employment,Growth, and Price
Levels for considerationby the Joint Economic Committee, 86 Cong. 1 sess. (1959),
pp. 44-77.
7. JamesTobin, "Inflationand Unemployment,"AmericanEconomicReview,vol. 62
(March1972),p. 9.
rationale rested on a model of "stochastic macro-equilibrium" in which

disequilibriain particular markets could maintain some stable positive in-
flation rate in the presence of some persistent degree of overall labor-market
tightness.
Asymmetries and nonlinearities could account for the puzzle of 1960-63
-a continuing rate of inflation of 1 or 2 percent in an excess-supply econ-
omy. If some prices and most wages are rigid on the down side and if the
response of wages to unemployment is nonlinear, then excess demand in a
minority of sectors can produce some upcreep in overall price levels.
Empirically, the Phillips curve looked like a winner for the United States.
Annual increases in prices (and wages) relative to unemployment rates
for 1954 to 1968 fit a hyperbola like a glove.8 A few refinementstook care
of the minor deviations. The persistence of the 1956-57 inflation into 1958
could be explained by some short lags of a vintage Keynesian type; the
somewhat better tradeoff performance of the early sixties relative to the
fifties could be accounted for by the price-wage guideposts or by "hidden
unemployment."9
The stubbornness of wage and price inflation in 1970-71 ended the hey-
day of the Phillips approach. Perry sought to save the Phillips curve by rein-
terpreting the measure of labor-market tightness to reflect the shifting
demographic composition of unemployment.10The Perry shift is now gen-
erally accepted as a constructive refinement, but it explains only a small
part of the "new" inflation of the 1970s. The expansion of unemployment
insurance and welfare benefits was also invoked to account for tighter labor
markets at a given unemployment rate.
The other major analytical developments stressed expectations. Accord-
ing to one thesis, the economy became more inflation prone because private
decisionmakers perceived increased tolerance for inflation and reduced
tolerance for unemployment on the part of government. I would not dis-
8. These gave no reason to suspect that the structurehad changedthrough 1968. In

termsof the relationshipbetweenpricesand the officialunemploymentrate,the observa-
tions for 1966-68 showed, if anything,a slightly more favorabletradeoffthan those of
1956-57. See the chart in the Economic Report of the President together with the Annual
Report of the Council of Economic Advisers, January 1969, p. 95.
9. GeorgeL. Perry,"Wagesand the Guideposts,"AmericanEconomicReview,vol. 57
(September1967),pp. 897-904; N. J. Simlerand AlfredTella, "LaborReservesand the
Phillips Curve," Review of Economics and Statistics, vol. 50 (February 1968), pp. 32-49.
10. George L. Perry,"ChangingLabor Marketsand Inflation,"BPEA, 3:1970, pp.
411-41.
ArthurM. Okun 355
miss this hypothesis." But I doubt that it explains much. For one thing, the
hypothesis would predict stronger spending behavior-a movement along
the Phillips curve, as well as a shift in it. If consumers and firms detect an
expansionary bias in stabilization policy, aggregate private demand should
be stronger for given settings of fiscal and monetary policies; I see no evi-
dence that this has been the case.
ACCELERATIONISM
Accelerationismwasthe most fundamental transformation of the Phillips

approach into an expectational format. It hypothesized that inflation will
become increasingly rapid in any maintained situation in which unemploy-
ment lies below some critical, or "natural," rate. Basically, it argued that
the very possibility of getting unemployment below the natural rate de-
pends on a process of fooling people-coaxing out higher employment and
higher production with higher prices for the things they sell and then sur-
prising them with higher prices than they expected on the things they buy.
Through lags in the perception of inflation, these surprisesraise output and
employment, but as people learn that they are being fooled, the lags shorten.
When Milton Friedman and Edmund Phelps independently set forth
this theory, the Phillips-curve approach seemed to be working very well.
Basically, accelerationismwas a pessimistic forecast ratherthan an explana-
tion of experience; whatever else one thinks of the theory, the prophetic
accuracy of its pessimism has to be admired.12Some of the macroeconomic
empirical facts of the early seventies fit the accelerationist theory. Even
though the unemployment rate exceeded the natural rate (by anybody's
estimate) in 1970-71, people were, according to the accelerationists, still
adapting to the inflationary surprises of 1965-69; hence, inflation deceler-
ated very slowly and only after a lag.
11. See Arthur M. Okun, "The Mirage of Steady Inflation,"BPEA, 2:1971, pp.
485-98, and othercontributionsto that symposium:WilliamFellner,"Phillips-typeAp-
proachor Acceleration?"pp. 469-83; and RobertJ. Gordon, "SteadyAnticipatedInfla-
tion: Mirage or Oasis?"pp. 499-510.
12. Milton Friedman,"The Role of MonetaryPolicy," AmericanEconiomicReview,
vol. 58 (March1968),pp. 7-11, and EdmundS. Phelps,"PhillipsCurves,Expectationsof
Inflation and Optimal UnemploymentOver Time," Economica,n.s., vol. 34 (August
1967),pp. 254-81. Actually, the essentialelementsof the theorywere spelled out much
earlierin William Fellner, "Demand Inflation,Cost Inflation,and CollectiveBargain-
ing," in Philip D. Bradley,ed., ThePuiblicStake in UnionPower(Universityof Virginia
Press, 1959).
More generally, the unemployment-inflation experience of the first half

of the 1970s manifestly reveals a far less favorable tradeoff than does that
of 1954-68. Clearly, the short-term Phillips curve has shifted upward. In
the sense of recognizing that shift, we are all accelerationists now (to re-
verse Friedman's celebrated concession to Keynes). On the other hand, I
believe the inflation-unemploymenttradeoff applies to all relevant inflation
and unemployment rates and all relevant time horizons. I find particularly
incredible the clear (though often ignored) implication of the no-tradeoff
view that inflation no longer imposes a cost. If the American public has
fully adapted to some anticipated inflation rate like 6 percent, then that in-
flation cannot do any good in expanding output and employment, but by
the same token it cannot do any harm in distorting distribution or alloca-
tion. The same public that allegedly adapted to that expected inflation rate
still believes that inflation is painful; and so do I.
The microanalytical underpinning of accelerationism is seriously defi-
cient. In part, inflation is supposed to distort temporarily the tradeoff be-
tween work and leisure. According to this story, when people observe a rise
in money wages, they believe that real wages are rising too. Consequently
they take jobs and give up leisure, which they now view as more expensive.
Ultimately, however, they find that the cost of living has accelerated too,
and the labor supply hence gradually shifts back. But why should people
take significantly longer to perceive the movement of the cost of living than
that of wages? Even more fundamentally, how can the thesis assume a
substantial positive elasticity of the supply of labor with respect to the real
wage? While that proposition has been widely accepted (by Keynes, among
others), the empirical evidence suggests that the elasticity is close to zero
and may not even be positive.13
In part, the temporary distortion is supposed to involve the tradeoff be-
tween work and search (rather than leisure).14According to this variant,
13. See Robert J. Gordon, "The WelfareCost of Higher Unemployment,"BPEA,

1:1973, table 2, p. 159, and George F. Break,"The Incidenceand EconomicEffects of
Taxation,"in Alan S. Blinderand others, The Economicsof PublicFinlance(Brookings
Institution,1974),pp. 180-91. Some demographicgroupsexhibit a positiveresponseto
their own wage, but offsettingthat are the negativecross-elasticitiesof the labor sup-
plied by one family memberto the wage of other members.
14. Models of work-leisure-searchdistortions include Dale T. Mortensen, "Job
Search,the Duration of Unemployment,and the Phillips Curve,"AmericanEconlomic
Review, vol. 60 (December 1970), pp. 847-62; Edmund S. Phelps, "Money-Wage
Dynamics and Labor-Market Equilibrium," Journal of Political Economy, vol. 76
ArthurM. Okun 357
becausethe unemployeddo not know the universeof job offers,they in-
terpreta good offerin an improvinglabormarketas a high relative wage
andtakeit ratherthancontinuingtheirsearch.Thus,the averageduration
of unemploymentand hencethe numberof unemployeddecline.But ulti-
matelyjob seekersrecognizethatit is worthwhileto takemoretimeto sort
the favorableoffersof a stronglabormarket.As a result,the dropin unem-
ploymentmust be temporary,lastingonly so long as peopleare fooled.
The facts of the real worlddo not squarewith the implicationsof this
work-searchhypothesis.First,as Perryhas shown,changesin the number
of spellsas well as in the durationof unemploymentaccountimportantly
for cyclicalfluctuationsin unemployment.Second,as Tobinhas noted,the
quitratedoesnot lag in an improvinglabormarket.Third,rejectionsofjob
offersare less frequentthan this theorywouldimply-except in the trivial
sense that everyunemployedpersonhas rejectedan "offer"to becomea
self-employedgarbagepicker.'5
Someaccelerationist modelsarerootedin productratherthanlabormar-
kets. In theirworld,outputexceedsits equilibriumlevel temporarilywhen
expectedsellingpricesare especiallyhigh in relationto expectedbuying
prices.'6I knowof no empiricalevidenceof suchcyclicallysystematicerrors
in firms'predictionsof sellingpricesrelativeto buyingprices.
Althoughtheirexplanationsare unsatisfactory,I believethat the search
of the "searchtheorists"has not beenin vain.Theyhavecorrectlyfocused
attentionon the waywagesandemploymentmustadjustthrougha search
processin the absenceof a market-clearing mechanismthat would adjust
wagesto keep supplyand demandin equilibrium.
The effortto buildthe analyticalbase for accelerationism,
led by Phelps,
has producedconstructiveresearchinto the "microeconomic foundations"
of bothemploymentandinflationtheory.Thepervasiveissueis the relative
role of quantityadjustmentsand of price(wage)adjustmentsin different
typesof marketsoverdifferenttimehorizons;this tiedbackinto the earlier
effortsby Hicks,Tobin, Patinkin,and Leijonhufvudto makesense of the
(July/August, 1968), pt. 2, pp. 678-711; and Robert E. Lucas, Jr., and Leonard A.
Rapping,"Real Wages,Employment,and Inflation,"in EdmundS. Phelps and others,
MicroeconomicFoundationsof EmploymentanidInflationTheory(Norton, 1970).
15. George L. Perry, "UnemploymentFlows in the U.S. Labor Market,"BPEA,
2:1972, pp. 263-75; Tobin, "Inflationand Unemployment,"p. 7; Gordon, "Welfare
Cost of HigherUnemployment."
16. See ThomasJ. Sargent,"RationalExpectations,the Real Rate of Interest,and the
Natural Rate of Unemployment,"BPEA, 2:1973, pp. 429-72, especiallyp. 435.
Keynesiantheoryof involuntaryunemployment.'7 And the commontheme

is that, in marketsthat lack a clearingmechanism,quantitiesvary a lot
becauseprices(wages)varytoo little and too late. Thesenonclearingphe-
nomenaare the heart of the explanationof both inflationaryand reces-
sionaryprocesses.
PURPOSE OF THIS PAPER
I shallnow sketcha nonclearingargumentthatmakessenseto me, draw-

ing on variousstrandsof the literature.Becauseof the absenceof market-
clearingmechanisms,quantityadjustmentscarry the burden for many
types of productand factormarkets,leadingto the observedsluggishness
and persistenceof inflationand of excessiveunemployment.While many
elements(includingsomemonopolymodels,transactionscosts, uncertain-
ties, and varioussunkcosts) can explainthe absenceof marketclearing,I
shall stressthe cost of information,interpretingit broadly(too broadly,
somemayfeel)to includecosts of prediction,of establishingreliability,and
the like.'8
In particular,the costsof informationleadto implicitlycontractuallong-
term relationshipsbetweenemployeesand employersand betweencus-
tomersandsuppliers.Theserelationshipscreatea zone of indeterminacy for
wagesandpricesanda needfor "fair"formulasfor the sharingof bilateral
monopolysurpluses.By puttingpriceand wagemakinginto a longer-term
context,theylengthenthe lags and weakenthe causalconnectionsbetween
changesin demandand changesin pricesor wages.
I shall then exploresome implicationsof the resultinglags in the infla-
tionaryprocess.Pricesand wagesin "customermarkets"will lag behind
17. John Hicks, CapitalandGrowth(OxfordUniversityPress, 1965),pp. 74-77; J. R.
Hicks, ValueanzdCapital(2d ed., London: Oxford UniversityPress, 1946),pp. 245-72;
JamesTobin, "Money Wage Rates and Employment,"in SeymourE. Harris,ed., Thle
New Economics(Knopf, 1947); Don Patinkin, Money, Interest, and Prices (2d ed.,
Harperand Row, 1965),pp. 313-34; Axel Leijonhufvud,On KeynesianEconomicsanld
the Economicsof Keynes:A Studyin MonetaryTheory(OxfordUniversityPress, 1968),
pp. 102-09.
18. I believemy semanticsare in the spiritof KennethJ. Arrow,"LimitedKnowledge
and EconomicAnalysis,"AmericanEconomicReview,vol. 64 (March 1974),pp. 1-10.
I do not mean that people are ignorant or generally uninformed;I share Michael
Wachter'sconcern (expressedin his comments below) that the terminology may be
misleadingin that respect.
ArthurM. Okun 359
those in "auctionmarkets."Thus,inflationmustaffectthe allocationof re-
sourcesand changerelativeprices.Hence,therecan be no uniform"real"
interestrate whose constancymight maintaina simple equilibriumof
financialmarketsduringinflation.
Customermarketsdependheavilyon-and in turnenhance-the useful-
ness of moneyas a yardstickand as a storeof value;that usefulnessis im-
pairedin a worldof inflation,as aremanyaspectsof buyer-sellerrelation-
shipsthat are "efficient"institutionsin a complexworld.Thus,the welfare
costs usuallyattributedto inflation-haphazardredistributionsof income
and wealth and socially unproductiveeconomizingof cash balances-
shouldbe viewedin a broadercontextas disturbancesto a set of institu-
tions that economizeon information,prediction,and transactioncosts
throughcontinuingbuyer-sellerrelationships.Inflationdoes fool people,
as the accelerationists
contend.But it does so, not so muchby disappoint-
ing theirpoint-estimateexpectationsas by deprivingthemof a way of eco-
nomic life in whichthey need not dependheavilyon the formulationof
costlyand uncertainpoint-estimateexpectations.
The skeletonI shallsketchbelowhas manymissinganalyticalbonesand
no empiricalflesh.Yet I believethis basicapproachhas far-reaching impli-
cationsfor both analysisand policy.It revealsweaknessesin proposalsfor
living with inflationby indexation.It explainswhy most of the American
publicdislikesinflationso muchmoreintenselythandoesmost of the eco-
nomics profession.It offersa common approachfor productand labor
markets,encompassingadministeredpricesand sticky wages.It suggests
that the welfarecosts of garden-variety, nagginginflationare qualitatively
similarto (althoughfar smallerthan)the distortionsof allocationtypically
attributedto galloping,acute hyperinflation.It explicitlyrecognizesthe
store-of-valuefunctionof moneyandrescuesinflationtheoryfrom"barter
illusion."This approachadvancesthe line of thinkingof my 1973 article
on upwardmobility,but contrastswith views I expressedas recentlyas
1970 and with most discussionsof inflationin the literature."9 The out-
19. "UpwardMobilityin a High-PressureEconomy,"BPEA, 1:1973,pp. 235-52; and

Arthur M. Okun, "Inflation:The Problemsand ProspectsBefore Us," in Arthur M.
Okun,HenryH. Fowler, and Milton Gilbert,Inflation:TheProblemsIt Createsand the
Policies It Requires(New York University Press, 1970), pp. 12-22. The best of the
"standard"treatmentsis Robert M. Solow, "The IntelligentCitizen'sGuide to Infla-
tion,"PublicInterest,no. 38 (Winter 1975),pp. 39-49.
standingexceptionis JohnHicks,whoserecentanalysisof the welfarecosts

of inflationdeliversthe sameprincipalmessageas does this paper.20
CustomerMarkets
In traditionalshort-runmarketanalysis,firmsareviewedas pricetakers
and quantitymakers.In the Marshalliancase, the prototypecompetitive
firmhas a givenquantityof freshfish or freshfruiton handthat it dumps
onto an auctionmarketto fetchwhateverpriceit will bring.2'Manifestly,
most sellersof productsin the real world are quantitytakersand price
makers;even those with minusculemarketsharesput pricetags on their
commodities.In the shortrun, they are neversurprisedby the price,and
alwayssubjectto surpriseabout the quantitiesthey sell.
If pricemakingmeantnothingmorethanthe absenceof auctionmarkets,
pricetags and pricelists wouldintroduceonly the tiniestlag in the adjust-
ment of pricesand trivialfluctuationsin quantities.Any firm(competitive
or monopolistic)thatexperienceda surprisingrisein demandwouldtendto
raiseits pricespromptly,andthoseunfavorablysurprisedwouldtendto re-
ducethem.A few otherconsiderationsmightlengthenthatlag a bit: it costs
somethingandtakessometimeto makeandimplementthe firm'sdecisions
on pricesand then to printand distributethe new pricelists. But theseele-
ments imply an inertiaof pricesthat should last for days or weeks, not
years.And they cannot explainmovementsin the wrong direction-like
risingpricesin the summerof 1975.
Aftera shiftin demand,competitiveproducerscould,in principle,reach
the new equilibriumpriceaftera few roundsof pricechangeswithoutan
auctioneer.Supposethereare many, fairlysimilarsmall hotels in a large
city, and that the marginalcost of havingan additionalroom occupiedis
negligible.With no auction(or organizedexchange)mechanismto keep
occupancyessentiallyfull andstableby varyingthe price,a downwardshift
in demandmight producea short intervalof depressedoccupancyrates;
20. See the concludingportion of The Crisis in KeynesianEconomics(Basic Books,
1974).Specifically(p. 79): "Any systemof prices [andwages]... is bound to work more
easilyif it is allowedto acquire,to some degree,the sanctionof custom.... This, I believe,
is the truereasonwhy inflationis damaging.... In conditionsof inflation[arrangements]
conltinuallyneed re-fixing,so that issueswhich had seemedclosed have to be reopened."
21. See AlfredMarshall,Principlesof Economics(8th ed., London: Macmillan,1947),
pp. 369-79, and Hicks, Capitaland Growth,pp. 49-57.
ArthurM. Okun 361
but the restorationof equilibriumwould not take very long, if the hotels
wereseriouslytryingto be do-it-yourselfauctioneers.No recessionwould
ensue,in whichthe hotels acceptless occupancyand maintaintheirroom
chargeswhileexpectingoccupancyto continuedepressed.Nor wouldpro-
longedbooms emerge.22
The absenceof the auctioneerhas more importanteffects.First, it con-
fronts the potential customerfor the hotel room with a costly search
process(like that of the unemployedworker).23 Even so, if the potential
customersampleda few hotels randomlyevery time he contemplateda
purchase,the lack of an auctioneerwouldmerelyreducethe priceelasticity
of demand,increasingthe effectivedegreeof monopolywithinthe industry
andthusthe profit-maximizing price.The realreasonthatthe auctioneeris
missedso muchand that hotelsdo not try to be do-it-yourselfauctioneers
is that the optimizinginnkeeperwill recognizehis abilityto influencethe
searchbehaviorof the customer.Today's occupantshave indicatedthat
they regardedthe hotel's offer(at least ex ante) as a satisfactorydeal. By
pledgingcontinuityof that offer,the innkeepercan encouragecustomersto
returnto buy, or at leastto sample,usingyesterday'sexperienceas a guide
to today's offerings.A kind of intertemporalcomparisonshoppingde-
velopsby whichyesterday'sofferinfluencestoday'sdemand,as a resultof
an impliedcommitmentof the sellerto maintainhis offer.
Thecustomerwho countson thatstabilitybelieveshe knowsthe termsof
his previoussupplier'sofferwithoutshopping.But he must shop to deter-
nmine the offers(price,quality,service)of unfamiliarsellers.It is as though
that informationis recordedon a card that can be purchasedfor a non-
trivialprice.In sucha situation,maximizingbehaviorforthe customermay
resemblesatisficingbehavior.Theexpectedvalueof the informationcardis
low whenthe statusquo is satisfactory.Knowingthis,the supplierwantsto
ensurethatthe customerwill not findit worthwhileto buy the information
card.He can assumethat, if today's offeris maintained,most of today's
customerswill return.If he also believesthat any rise in his pricewill in-
22. Nor is it conceivablethat, despitethe shift in demand,each hotel ownerbelieves

thatthe priceelasticityof demandat the prevailingprice,and hencethe profit-maximizing
price,is unaffected.
23. See the model presentedby EdmundS. Phelps and SidneyG. Winter,Jr., "Opti-
mal Price Policy under Atomistic Competition,"in Phelps and others, Microeconomic
Foundations.I am introducingone new elementthat is not in their model: relatingthe
customer'ssearchbehaviorto the supplier'spricingstrategy.
creasediscretelythe proportionof his customerswho will buy information

cards,he will tend to leave his priceunchanged,even whendemandrises.
THE BUYER-SELLER RELATIONSHIP
An establishedcustomer-supplier relationshipintroducesa bilateralmo-

nopoly surplus that can be split between the cooperatingbuyers and
sellers.In the shortrun,mostcustomerswouldpaya slightlyhigherpriceto
theirsupplierswithoutshopping,and most supplierswould,if theyhad to,
sellfora shadelessto theircustomers.Thisinterdependence putsa premium
on maintainingthe relationshipand on limitingconflictoverthe sharingof
the surplusto methodsthat will not impairits total value.
The supplierobviouslycannotpromisethe customerthathe will offerthe
samedeal forevermore.In particular,he may have to raisehis priceif his
costs rise. But he can promiseto treatthe customer"fairly"on all the di-
mensionsof theirtransactions,thus offeringthe customeran implicitcon-
tract.It remainsimplicitbecauseof the highcost of spellingout and nego-
tiatingthe termsof an explicit,formalcontract.The implicitcontractmay
apply to a wide variety of unspecifiedcontingenciesextendingbeyond
changesin costs.Whenthe supplierandcustomeragreeon certainrulesor
conventionsof fairplay,each offersthe otheran effectiveincentiveto play
by those rules(and a crediblethreatagainstthe one who breaksthem).24
Theseimplicitcontractsserveas efficientsubstitutesfor the costly institu-
tions of formallong-termcontractsand organizedfuturesmarkets.25
Empirically,the typicalstandardof fairnessinvolvescost-orientedpric-
ing with a markup.Apparently,in most industries,the criterionis full
ratherthanvariablecosts. But it is basedon somestandardor normalout-
putratherthanactualoutput;thusthe customeris not askedto pay for the
higheroverheadperunitandthe lowerproductivityof recessions.The con-
ceptis also apparentlyhistoricalcosts,whichare obviouslysubjectto more
precisecalculationthan are replacementcosts. The recurrenceof transac-
24. For a discussionfocusingon the labormarketthat considersa varietyof informal

methods of dealingwith contingencies,see OliverE. Williamson,MichaelL. Wachter,
and Jeffrey E. Harris, "Understandingthe Employment Relation: The Analysis of
IdiosyncraticExchange,"Bell Journalof Economics,vol. 6 (Spring1975),pp. 250-78.
25. For the problems associated with futures markets, see Arrow, "Limited
Knowledge."
ArthurM. Okun 363
tions betweencustomersand suppliersmakesit possibleto base the terms
of saleon the factsof the pastratherthanforecastsof the future.Oncesuch
cost-basedpricingpracticesbecomeestablished,the firmcan often count
on its competitorsto set pricesin a similarfashion.2"
It can justify cost-orientedpriceincreases-a desireevidentin the dedi-
cated,if fuzzy,statementsthatfirmsissue,insistingthathighercosts "force"
them to raise prices.No suppliercan tell his customers:"As a result of
strongerdemand,I am now in a positionto capturea largershareof the
surplusfrom our relationship."This attitudeinfluencesprice-makingbe-
havioras well as public-relations releases.In effect,the supplierfirmrepre-
sents itself to its customersas a kind of procurementagency operating
undera brokeragearrangement.The markuponto costs becomesa rea-
sonableway to set a "fair"pricefor the servicesof the firm.
More generally,the customer-strategy entrepreneur gearspricesto costs
becausehis relianceon customersvastly complicatesany fine-tuningof
pricesto demandfor purposesof profitmaximization.The priceelasticity
of demandfor his productis relativelylow in the shortrun becauseof his
establishedclientele(and that of his competitors),but muchgreaterin the
long runbecauseinformationdiffuses.Hence,thefirmpresumably"under-
prices"-or sacrifices-for the near term if it maximizesthe discounted
futurestreamof profits(ratherthan current-period profits).Thus, a price
reductionin responseto a temporaryweakeningof demandwould often
imposean even greatersacrificeof currentprofitsin orderto promotefu-
tureprofits.Moreover,the cruciallong-runpriceelasticitiesof demandare
subjectto highdegreesof riskor uncertainty,or to estimationonlythrough
expensivemarketresearch(that is, high informationcosts). A risk-averse
firmmay be discouragedtherebyfromchangingpricesbecauseit sees high
variancein the outcome(a'la Brainard)or becauseit is led to "slant"the
probabilities(a'la Fellner).27In addition,risk-averseinvestorsmaylike the
reducedcyclicalvolatilityof profitsthatresultsbecausefirms"underprice"
leastin periodsof weakdemandand most (thus,investingmost for future
demand)in particularlygood times.
26. PhillipCagan,TheHydra-HeadedMonster:TheProblemof Inflationin the United

States (AmericanEnterpriseInstitute, 1974),pp. 21-34.
27. WilliamBrainard,"Uncertaintyand the Effectivenessof Policy," AmericanEco-
nomicReview,vol. 57 (May 1967), pp. 411-25; WilliamFellner,Probabilityand Profit
(Irwin, 1965),pp. 173-80.
THE SCOPE OF CUSTOMER RELATIONS
Customerrelationshipshavesomeimportanceto anyfirmwhosedemand
curveis higherthanit wouldbe if the firmwereofferingits productfor sale
for the firsttime-with no clientele,reputation,or recordof performance.
Thoserelationshipsshouldbe most importantfor heterogeneousproducts
and least importantfor homogeneousones, wherethe price is uncompli-
catedby a qualitydimension.Nonetheless,heterogeneityhas manyaspects
beyondthe physicalcharacteristics of the product,includingtransportation
arrangements, creditterms,speedand reliabilityof delivery.Thus, depart-
ment storesfacilitatethe returnof merchandisefor full refundby regular
customerswith chargeaccounts.Wholesalerswho have establishedthe re-
liabilityof smallretailersfor tradecreditofferfinancingarrangementsto
keep them as customers.Similarly,whilecustomerrelationshipsseemless
importantfor professionalbuyersthan for households,even the former
cannotassessthe long-termreliabilityof a new supplier,and henceprefer
continuingrelationships.Thus,in a weaksteelmarket,importedsteelmay
be priced 10 percent or more below the physicallyidentical domestic
product.In a strongmarket,however,the reliablecustomerof domestic
firmsis assuredof supplies,in amountsgearedto his past purchases,at
pricesbelow those of imports.
Big-ticketitems that are bought infrequentlymight appearto be least
subjectto the customerstrategy.Yet, repairserviceson autos and appli-
ancesareusedto maintainrelationships;andfirmsworkto establishbrand-
name reliability,in effect countingon reputation(a flow of information
from one consumerto the next) to substitutefor repetition.Finally, oli-
gopolisticmarketstructuresthat make price competitionself-destructive
amongfirmsmay encourageuniformand sticky"administered prices"ac-
companiedby differentiation of services.28
28. It is thus understandablethat GardinerMeans and John Blair have found and
stresseda correlationbetween industrial-concentration ratios and the kinds of pricing
patternsthat I attributeto customermarkets.See National ResourcesCommittee,The
Structureof the AmericanEconomy,A Report Preparedby the IndustrialSection under
the Directionof GardinerC. Means (GovernmentPrintingOffice,1939),pt. 1, pp. 138-
45; GardinerC. Means, "SimultaneousInflationand Unemployment:A Challengeto
Theoryand Policy," Challenge,vol. 18 (September/October1975),pp. 6-20; and John
M. Blair, "MarketPower and Inflation:A Short-RunTargetReturn Model," Journal
of Econiomic Issues,vol. 8 (June 1974),pp. 453-78. I do not, however,believethat sticky
cost-basedprices are unique to oligopoly or could be eliminatedby changing market
structure.
ArthurM. Okun 365
In summary,the prototypeof pricingin customermarketsis a markup
overpastcosts. But cost changesshowup with a lag becausepriceswill be
alteredonlywhenaveragecostshavechangedby somethresholdamountor
only at specifiedintervals.As Cagan summarizesthe result: "Empirical
studieshave long found that short-runshifts in demandhave small and
often insignificanteffects[on prices],and that, instead,costs play a domi-
nant role."29This empiricalfindingof markuprigidityis inconsistentwith
the classicaltheoryof pricedetermination, in whichthe strengthof demand
shouldalterthe ratio of pricesto costs.
CYCLICAL PATTERNS
By dampeningits cyclicalpricefluctuations,the customer-strategy firm

will magnifyits cyclicalvariationsin ordersand sales. Whenthe firmen-
countersan unexpecteddrop in demand,its inventoriesof finishedgoods
may increase;its utilizationof its work force and capitalmay decline;it
may cut its workweek;it may reducehiringand beginlayoffs; or it may
speed up deliveries.And these changesin quantitiesmay not be accom-
paniedby changesin price,for the reasonsspelledout above.
When demandstrengthensafter a slump,the same optionsare thrown
into reverse.The firmpresumablyoptimizesso that it is equallycostly to
meet favorablesales surprises(includingsuccessesin attractingnew cus-
tomers)by dippinginto inventories,lengtheningits workweek,anddrawing
on its queueof job applicants.Undertypicalcircumstances, the customer-
strategyfirmis likelyto maintainsomereserveof capacityin all threeforms.
The firmmay be in disequilibrium throughoutand yet makeall the adjust-
ments in quantities.The forces inducingchangesin prices can be very
weak.30
29. Cagan, Hydra-HeadedMonister, p. 22; also see the detailedreview of empirical
findingsby WilliamD. Nordhaus,"Recent Developmentsin Price Dynamics,"in Otto
Eckstein, ed., The Econometrics of Price Determiniation, A Conference sponsored by the
Board of Governorsof the FederalReserveSystemand Social ScienceResearchCenter
(Boardof Governors,1972),pp. 34-43. Moreover,the analyticaldiscussionof short-run
pricingby Nordhauson pp. 21-27 and 31-34 is particularlyperceptive.Anothervaluable
discussion is presented by Otto Eckstein, "A Theory of the Wage-PriceProcess in
Modern Industry,"Review of EconomicStudies, vol. 31 (October 1964), pp. 269-71,
281-82. For an analysis of the rationaleof fixed pricingintervals,see StephenA. Ross
and Michael L. Wachter, "Pricing and Timing Decisions in Oligopoly Industries,"
Quarterly Joulrnal of Ecoiionoics, vol. 89 (February1975), pp. 115-37.
30. In Vie customer-strategymodel, cost decreasesare no less likely to be passed
throughin lower pricesthan are cost increasesof equalmagnitudein higherprices.This
366 BrookingsPaperson EconomicActivity,2:1975
Career Labor Markets
Customer-supplier relationshipsin productmarketsare paralleledby

long-termemployer-employee attachmentsin labormarkets;in combina-
tion, the two modelsunify the analysisof quantity-adjusting responsesto
disequilibrium in labor and productmarkets.
Actually,the labor-marketsideof this analysisis betterknownandmore
widelyaccepted,and since I have previouslyexploredit in some detail,I
shalltreatit only brieflyhere.3'Manyemployersadoptpoliciesto promote
long-runattachmentby workers.They pursuethat strategybecausethe
valueof an experiencedworkerexceedsthat of an inexperiencedone by a
margingreaterthanthe correspondingwagedifferential.Enteringworkers
impose costs on the firm of screening,hiring, training,and on-the-job
learning;yet they cannot be made to bear these costs fully through a
loweredenteringwage. For one thing, many of these "investments"are
valuableto the workeronlyinsofaras he remainsan employeeof thatfirm;
the nontenuredrecruitwili not pay for them. Second,the relevantinterest
rateto the workeris likelyto be farhigherthanthatto the firm.Third,even
the "general"trainingthat improvesthe worker'smarketabilityto other
firmscannotbe properlyappraisedby him, givenhis information.Finally,
the firmwill wish to bolsterthe enteringwage in orderto ensurea queue
of willingapplicantsthat it can sort and screenwhenit needsto hire new
workers.
Thus, the firm reallymakes an investmentin a new worker,spending
moreto hire, train,and pay him than he is worthin the shortrun. But it
mustthen amortizethat investmentby payingwagesbelowmarginalreve-
nue productto workerswho become experienced.Nonetheless,the firm
wantsto maintainwageratesfor experiencedworkersabovethose of their
next-bestopportunityin the labormarketin orderto hold downquitrates
andprotectits investment.It will also hold down quit ratesby offeringits
resultis consistentwith Tobin's view of the empiricalevidence:"This [inflationary]bias
cannotbe attributedto productpricing,which apparentlypasses on proportionatelythe
changesin labor costs . .. in both directions-down as well as up." See James Tobin,
"The Wage-PriceMechanism:Overviewof the Conference,"in Eckstein, ed., Econ-
ometricsof Price Determination,p. 10.
31. Okun,"UpwardMobility,"pp. 235-44; also, see StephenA. Ross and MichaelL.
Wachter,"Wage Determination, Inflation, and the Industrial Structure,"American
EconomicReview,vol. 63 (September1973),pp. 675-92.
Arthur M. Okun 367
experiencedworkersregularpay increasesand growingseniorityprivileges

in the form of fringe benefits,retirementprograms,and provisionsfor
leave and vacation.These arrangementscreatea bilateralmonopolysur-
plus and a zone of indeterminacyfor the wagefor an experiencedworker.
That wagemust exceedhis perceivedbest alternativeand be less than his
currentmarginalrevenueproduct.Again, because of informationcosts
(andsunkcosts),the elasticitywithrespectto the wagehas a time gradient
-from low in the short run to high in the long run. These relationships
have pervasiveempiricalconsequences-the negativecorrelationbetween
wageratesandquitrates,theextremelylow quitrateof experiencedworkers
in high-payingindustries,the countercyclicalchangesin the size of wage
differentials,cyclicalpatternsof upgradingjobs, the low levels of unfilled
vacanciesfor careerjobs even in tight labor markets,and the like.
Thislong-attachment, or "career,"job marketis analogousin manyfea-
turesto the customer-product market.But the roles playedby buyersand
sellerspresenta set of interestingcontrasts.The price(wagerate)is set by
the buyerin labor markets(outsideof collectivebargaining),but by the
seller in product markets.The other party then engages in search (or
shopping).A sellerin productmarketsand a buyerin labormarketstypi-
cally make similartransactionswith manypeopleon the otherside of the
market,andtheymustbe pricemakers,at leastin part,to displayequitable
treatmentof the peoplewith whomthey deal.Wagediscriminationamong
similaremployeesof the samefirmor pricediscriminationamongits cus-
tomerswould destroylong-termrelationships.
CYCLICAL RESPONSES
Careerjob arrangements desensitizethe wagerateto changesin overall

tightnessof the labor market.First, becauseexperiencedworkersearn
morethan theirnext-bestopportunitywage, strengtheningthe labormar-
ket willincreasethe quitrateonlyslightly.Second,the normalqueueof ap-
plicantsallows the firmto hire new workerswithoutraisingthe entering
wage.To be sure,the "marginaleffectivewage"of the firmmay increase.32
A higherrate of inflowof new workersmay push up marginalhiringand
trainingcosts and forcethe firmto hire new workerswith poorercreden-
tials. Hence,it may be worthwhileto lengthenthe workweek,even at the
32. Robert E. Hall, "The Process of Inflationin the Labor Market,"BPEA, 2:1974,
pp. 347-60.
cost of payingovertimepremiums.But the firmhas a strongincentiveto

avoidan unscheduledincreasein the enteringwageinsofaras it wouldfeel
obligedto maintainits wagestructureby increasingsimultaneouslythe pay
of experiencedworkers.
By makingthe elasticityof laborsupplylow in the shortrun and uncer-
tainin the long run,the nonauctionsearchmechanismcreateswageinertia
in careerlabormarketsjust as it makespricesstickyin customerproduct
markets.Moreover,the quit-ratefunctionmay discouragewage changes
muchas the "informationcard"decisiondiscouragespricechanges.That
aspectof wagestickinesscan be illustratedby the followingmodel.A firm
hastwo classesof workers-experiencedand new; theirwagedifferentialis
fixedby customor convention.Thus, the firmhas a singlewage decision;
that,in turn,will determinethe quitrateof experiencedworkers,andhence
the residualnumberof themwho stickwiththeirjobs. So long as the firm's
optimalemploymentexceedsthat level, it will be hiringsome new workers
out of its queueof applicants,whosequality(innateskill, education,expe-
rience)may be assumedto have a distinctand identifiablegradient.For
any givenwage,the deeperthe firmmust dip into the queue,the lowerwill
be the productivity(averageand marginal)of new workers.Thus, the
qualityof newworkerswilldependpositivelyon the wageandnegativelyon
the level of employment.
By settinga higherwage,the firmgainsby reducingthe quit rate of ex-
periencedworkersand by raisingthe qualityof its queue.It pays for that,
of course,with a biggerpayroll.If the supplyof new applicantsof given
qualityand the quit rate of experiencedworkerswere both smooth con-
tinuousfunctionsof the wage, the optimumwage would turn out to be a
continuousincreasingfunctionof the level of employment.But the quit
rate of experiencedworkersis likely to be discontinuous-low and fairly
insensitivefor wages at or above the level that fulfillsthe firm'simplicit
long-termobligationto be fair,butjumpingdiscretelyfor wagesbelowthat
level. If the discontinuityaroundthe "fair"wage is substantial,that wage
may be optimalfor a wide rangeof employmentlevels.
In the event of a suddendeteriorationof the labormarket,wagereduc-
tions maynot playmuchof a role (even for a nonunionemployer).A firm
that has sought to establisha prospectof rising wages over the career
horizonmaydestroyits investmentby cuttingwages.To cut wagesand yet
avoid a long-termworseningof the quit rate, the firmmust prove to the
experiencedworkerthat it is "forced"to take such action,whichmay be
possibleonly if the firmis on the brinkof bankruptcyor at least of a shut-
ArthurM. Okun 369
down.Otherwise,pay cuts could be interpretedby workersas an "unfair"
action by the employer to capture a larger share of the surplus.33
Whenthe alternativesare temporarylayoffsand wage cuts, the former
are likelyto be the lesserevil to the firm.To be sure,they are an evil: ex-
periencedworkerscareaboutsteadinessof employmentas weli as the wage
rate, and a laid-offworkermay find a betterjob beforehe is recalled.In
fact, however,wage premiumsin many industriesare sufficientlylargeto
bringmost laid-offworkersback when they are recalled.34 Moreover,as
Baily has recentlypointed out, layoffs have severalpartialoffsets in the
form of transferincomeand some value of leisure.35 Finally,and I believe
particularlyimportant,the firm has clean hands in the case of a layoff;
since it is not using the worker,it cannot be "takingadvantage"of him.
All in all, it is easy to understandhow the conventionthat wagesare cut
only in extremis becameentrenchedin labor marketsbeforeunions were
significantand beforeKeynesattachedso much weightto the downward
rigidity.
For many of the same reasons,a weakeningof labor marketsmay not
even substantiallyreducethe short-runrateof wageincreasein careerjobs.
The big issuein wagesettingis a long-runbattleoverthe divisionof the bi-
lateral monopoly surplusbetween employersand experiencedworkers.
Each side perceivesthe short-runelasticityof the other as relativelylow,
but the long-runelasticityas substantial.Hence, workersdo not want to
squeezethe firm into gradualextinction;and the firm does not want to
press the workersto the point of explosivequit rates.These boundaries,
however,may containa wide area of sharpconflictof interest;and stan-
dards of fairnessare sought to preservethe surplusavailableto the two
sides combined.
CRITERIA IN WAGE DETERMINATION
Variousgroupsof employersandworkersmayopt fordifferentstandards

of "fairness":gearingwagesto otherwagesor to productprices,or to con-
33. Some case studiesof the extremeconditionsthatjustify pay cuts are presentedby
Peter Henle, "Reverse Collective Bargaining?A Look at Some Union Concession
Situations,"Industrialand LaborRelationisReview,vol. 26 (April 1973), pp. 956-68.
34. HaroldL. Sheppardand A. HarveyBelitsky,TheJob Hunt(JohnsHopkinsPress,
1966),pp. 35-37.
35. MartinNeil Baily, "Wagesand EmploymentunderUncertainDemand," Review
of EconomicStudies, vol. 41 (January1974),pp. 43-47. The psychic cost of unemploy-
ment to most breadwinnersis an offset to those offsets, however.
sumerprices.Any of these standardswill weakenthe impactof changing
labor-marketconditions,and strengthenthe momentumof a givenrate of
wageincrease.
Wage-wage patterns.One methodof achieving"fairness"involvesemu-
latingthe wagesrecentlyset by otherfirms.The comparabilitysystemfor
compensationof federalemployeesis the most importantexampleof this
practice,but similarmethodsare used in many privatefirms.And many
smallerindustries"pattern"on steel or autosin theircollective-bargaining
settlements.Suchpatterninglendsinertiato the rateof wageincreasein the
economy.In particular,overlappingschedulesof contractsof long duration
may resultin a wage settlementthat is a moving averageof other recent
wage settlementsthat were in turn moving averagesof previousmoving
averages.
Wage-productprice patterns. Alternatively, long-attachment wages may
followproductprices.Recentpricehikesby the employerthat widenedhis
profitmargincan be legitimatelyviewedby laboras increasingits marginal
revenueproductand hencewarrantinga pay advance.On the otherhand,
recentrisesin the firm'spricethat merelyreflecteda passthroughof previ-
ous increasesin laboror materialcosts,withan unchangedmarkup,should
make no case for furtherwage gains.36
Wages and consumer prices. Finally, consumer prices may influence
wages.Presumably,one of the attractionsof a careerjob is the prospectof
a reasonablysecureand risingstandardof living for the long run. If the
cost of livingincreasessharplyrelativeto wages(or if the real-wageoutlook
merelybecomesfar more uncertain),the workermay see less value to his
job, perhapsenoughless to quit, even with no evidencethat his wagehas
deterioratedrelativeto the (unknown)alternatives.And if the cost of living
would affectquit rates at given wages, it will influencewages-in a way
that need not dependon a mythicalreal-wageeffecton leisure.
If the risk of a price-wagesqueezeimposes a serious disutilityon the
worker,unionswill seekwageadjustmentsinformallygearedto pastmove-
36. Many wage equationsin the early sixties included the profitmarginas an inde-
pendent variable. See Richard G. Lipsey and M. D. Steuer, "The Relation between
Profitsand Wage Rates," Economica,n.s., vol. 28 (May 1961), pp. 137-55; Otto Eck-
stein and ThomasA. Wilson,"The Determinationof Money Wagesin AmericanIndus-
try," QuarterlyJournalof Economics,vol. 76 (August 1962), pp. 388-401; George L.
Perry,Unemployment, Money WageRates, andIniflation (M.I.T. Press, 1966),pp. 27-29,
48-52. The profitsqueeze and wage bulge of subsequentyearsmade the fit deteriorate,
so the variabledroppedout-but not for analyticalreasons.In part, Perryputs it back
in his discussionof income sharesin the articlein this issue.
ArthurM. Okun 371
ments of consumer prices or formal escalator clauses (even though an opti-
mal bargainingstrategy would depend on the ability to pay of the employer
as well as the ability to consume of the worker).
On reasonable assumptions, the worker should prefer a contract for the
next three years that offers certainty about his real wage to one with the
same expected value that provides a certain nominal wage (and hence an
uncertain real wage). On standard assumptions of one-sector inflation
models, the employer who must commit himself for three years would also
be expected to make the same choice. Surely, the wage escalator should re-
duce risk for firms in cyclical industries where product demand (and hence
the ability to pass increased labor costs into product prices) is typically
highly correlated with movements of overall consumer prices. The fact that
formal escalators are the exception rather than the rule reveals a defect in
the standard theory. Why do employers uniformly resist escalators, even in
industry-wide bargaining?In a few conversations I have had with sophisti-
cated businessmen on this issue, their explanations illustrate vividly the
importance of certainty about nominal future costs in customer product
markets. Some mention the need to make fixed-pricebids or to accept fixed-
price orders for production processes of long duration. More generally,
they insist on the importance of planning in terms of dollars. They stress
vulnerability to interproduct and international competition in a world
where escalator clauses are rare. They view their future financial resources
as given in nominal, and not constant, dollars. When they have to decide
on a purchase of labor-saving machinery, they want to compare its price
with some known price for labor over as long a horizon as possible.
Even if, for such reasons, the escalator imposes a net cost on the em-
ployer, the firm may recover that cost with an offer of lower expected value
if the insurance is worth a lot to the worker. Thus, escalators may be an
efficient element in the wage bargain. In fact, they have spread in recent
years, although really strong uncapped ones are limited to a few major in-
dustries. All in all, consumer prices influence wages through both formal
escalators and informal cost-of-living adjustments-ruling out any pure
case of a "one-shot" rise in the cost of living.
INFLATIONARY BIAS IN WAGES
Of the various features of a career labor market outlined above, only the
downward rigidity of the wage level points to an inflationarybias in wages.
Because a wage cut is a qualitatively different and abnormal outcome, it
can have sufficientlyadverse long-term effects on quit rates to become un-
profitable for employers. The qualitative significance of zero as a floor has
no parallel for any ceiling. While downward rigidity in this sense can im-
portantly raise the mean wage increase if much of the frequencydistribution
of wage changes would otherwise lie below zero, it cannot have any influ-
ence when the entire frequency distribution lies in the positive range.
Apart from downward rigidity, an inflationary bias might arise from an
asymmetry following a regularly scheduled wage increase or a collective-
bargaining settlement. The firm still retains the option of "overfulfilling"
its pledge, by awarding extra wage increases or Christmas bonuses if labor
markets should tighten dramatically. But it cannot underfulfill with nega-
tive Christmasbonuses. The resulting phenomenon of upward "wage drift"
introduces a long-term bias if any of the temporary bulge becomes perma-
nent (as seems likely, if only because it gets into prices in the interim).
A third possible source of inflationary bias arises because the strike is a
recurrent threat that may disrupt any static equilibrium for union wages.
Analytical models typically suggest that, while monopoly power of unions
may lead to high wages, that power, once exercised, should not speed the
rise of wages over time. But because the strike threat is the enforcement in-
strument of the union's (bilateral) monopoly power, that standard conclu-
sion is not airtight. In any formal contract, the union makes the concession
of disarming itself and burying the strike threat for, say, the three-year
duration. When the contract expires, it once again has the opportunity to
exercise that threat; conceivably, it may seek an additional payment from
the firm in return for three more years of disarmament.37
In summary, career labor markets make wages less sensitive to labor-

market conditions, rigid downward, and subject to a zone of indeterminacy
and possibly to an inflationary bias. They also create indeterminacy in the
bargainingprocess, permitting widely varying results in differentperiods of
history and different countries on the extent to which wages follow labor-
market tightness, wages, product prices, or consumer prices. Some charac-
teristics of wage behavior are predictable: the rate of wage increase will
have momentum; it will ultimately respond to market conditions; it will
show downward rigidity. Once adopted, a particular pattern of wage fol-
37. This argumentdoes not imply that the union will capturean ever-largershare of
the bilateralsurplusin successiveroundsof bargaining.For one thing,marketconditions
do change;for another,firmswould ultimatelyfind it worthwhileto resist furtherwage
increaseseven at the expenseof greaterrisk of a strike.
Arthur M. Okun 373
lowing will persist for some time. But it will change and evolve, and it will
defy any generaltheory with empirical content that is meant to comprehend
all labor markets in all countries at all times-as is confirmed by Perry's
analysis of worldwide wage experience in this issue.38
Inflationin the Mixed Customer-AuctionEconomy
In the real world, a few products are sold on organized auction markets;
a few strictly fit the customer-marketmodel with price tags based on a non-
cyclical markup over past costs; many lie between these poles; and some
may reflect forces that are neither market clearing nor customer oriented.
While no labor is marketed in a pure auction exchange, labor markets
cover a spectrum of wage flexibility. In this sense as well as in many others,
it is a mixed economy.
Figure 1 illustrates the mix in product markets, showing the contrasting
behavior of four price indexes in selected periods of varying inflationary
pressure. The sensitive industrial materials index, bar D, approximates
pure auction prices; it is volatile, often declines, and responds promptly to
changes in aggregate demand pressure. A broader group of wholesale in-
dustrial materials, whose price changes are shown by bar C, is not quite so
volatile, less apt to fall, and less prompt to respond. (The unusual relative
behavior of bars C and D over the period 1973:4 to 1974:4 stems from the
inclusion of petroleum in C, but not in D.) Wholesale prices for finished
products (bar B) are less "auction-like" in behavior than either C or D,
while the private nonfarm deflator (bar A) is the least auction-like and
most dominated by customer prices.
A SIMPLE ALGEBRAIC MODEL
A simple algebraic model can describe some of the mechanisms of infla-

tion in the mixed economy. For convenience, I shall focus on the auction-
38. Throughout,some of the phenomenathat I view as conventionalcould be con-
ceptualizedas expectational.(The indeterminacyis there in either case.) In part, this
choice is a matterof analyticaltaste. For example,I would describethe currentsalaries
of baseball playersas based on last year's performanceand argue that this convention
rests on the long-termattachmentthat providesan incentiveand a payofffor everyyear
(exceptthe finalone in eachcareer).Chicagoand Minnesotaeconomistswouldinsistthat
the pay is an expectedvalue wherelast year'sperformanceprovidesthe best estimateof
this year's.As I see it, the analyticalprincipleof Occam'srazorsupportsmy preference,
but perhapsan even more ancient principleapplies: de gustibusnon est disputandum.
o~~~~~~~~~~~ -
0 ~~~~~~~~~~~~~~~~<~~~~0
2~~~~~~~~~~~~~~~~~~~~~~~~~~,
I I' l-
rA ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ r
.>~~~~~~~~~~~~~~~~~e
eD 2N
C.)
cnXI
R2 > ...
0 ~~~~~~~~~~~~~~~~~...... .
mP.* 1 1
0MX2 t
0.
~~~~~~r-
N b:
wlv ~~~~~~~~~~- N
- _- - r~~~~~~~~~~~~~t
0
.0
N N o
.0)
CU3
eM -
~ ~ ~ SM
0
C-.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I ..c~.. Cto.
C
< * _ W~~~,> 5 c0.
o o
t u
$X
'~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1
o o
3aeg~~~~~~~~~~
E
CU0
customerdichotomyin productmarketsratherthanthat in labormarkets.
I shall assumethat customerpricesare set by a markupover cost with a
one-periodlag, whileauctionpricesclearmarkets.I shallalso assumethat
all the costs of customerproducts(and none of the costs of auctionitems)
are wages.For simplicity,the supplyof auction-priceditems is taken as
completelyinelasticoversomesignificantshortperiod,whilethe outputof
customeritemscan be expandedat constantreal costs. Moreover,I shall
initiallyignoreany changesin stocks of auctionitems (treatingthem all
like freshfish).
Also for simplicity,incomeand priceelasticitiesof demandare takenas
unityfor bothtypesof products,so thateachcapturesa fixedshareof total
expenditures.For the moment,I shall note merelythat wages, W, may
dependupon prices,P, in both sectors,output, Q, in the customersector
(whichreflectsthe strengthof demand),andlaggedwages,W-1.The model
can be representedby the followingfour equations:
(1) PAQA = aY
(2) PCQC (1 - a)7

(3) PC = bW1
(4) W = f(PA ,PC, Qc, W1),
wherethe A subscriptrefersto auctionitems,and the C subscriptto cus-
tomeritems; F is nominalincome,which(like QA)is taken as exogenous.
Initially,supposethat pricesin both sectorsare constantand so is in-
come.Now incomestartsrisingpersistentlyat a rateof m percent,perhaps
becauseof stimulativemonetaryor fiscalpolicies.The auction-priceditems
immediatelyhave an inflationrate of m. But the inflationrate of customer
itemsinitiallyis zero, whiletheir outputexpandsat the rate of m.
Developmentsin subsequentperiodsdependon the behaviorof wages.
In the extremecase wherewagesare absolutelyrigid,
(4a) W = constant,
the inflationin the auctionsectorand the expansionof outputin the cus-

tomer'ssectorwould both continueat the same rate of m. The inflation
wouldneverspreadto the customersector,and real wage rateswould be
squeezedalthoughreal payrollswouldbe maintained.
Supposeinsteadthatwagesrespondpromptlybutincompletelyto output
ArthurM. Okun 377
(and henceemploymentdemand)in the customersector,but not to prices:
(4b) W/W-1 = (QC/QC),9
whereQCis the initial"noninflationary level"and0 < s < 1. In this case,
Pc startsrisingafter one periodand keepsrisingat an increasinglyrapid
ratethat ultimatelyreachesm and thus matchesthe rate of increaseof PA.
But the ultimateratio of PC/PAis below its initialvalue by m/s. And the
equilibriumvalueof log (Qc/ QC)is alsom/s. Finally,the equilibriumvalue
of PC/Wis m below the initialratio.Thus,real productwagesare up and
profitmarginson customeritemsare down.Real wagesmeasuredin terms
of consumerpricesmay,however,be up or downdependingon the weight
of auctionitemsin the cost of living.Thus,the initialinflationaryphasein
whichauctionpricesrise fasterthan customerprices(and wages)"perma-
nently"changesrelativeprices.39
Finally, suppose wages iespond fully and immediatelyto the cost of
living, but not at all to the state of the labormarket:
(4c) W/WO = (PA/PAO)k(PC/PC 0)l,
wherek is the weightof auctionitemsin the cost of living.It is convenient

to set PAO= PcO= bWo = 1. Remarkably, every single result of this model
correspondsto thatbasedon (4b),withthe meresubstitutionof k fors. The
(4c)variantdoesmakerealwagesin termsof consumerpricesdeterminately
constant,but the other changesin relativepricesthat emergefrom (4b)
applyhere as well.
Withboth (4b) and (4c), endingthe growthof nominalincomedoes not
halt inflationpromptly,althoughauctionpricesstop inflatingat once.Cus-
tomerpricesstill registerpast wageincreases,whilewageskeeprisingas a
resultof tight labor markets-in (4b)-or of continuingadvancesin cus-
tomerprices-in (4c). Thus, inflationdampsdown only gradually.
Interestingfurtherpossibilitiescan be entertainedby relaxingsome of
the restrictiveassumptionsmade above and thus allowingfor more com-
plexlags,partialresponsesof wagesto the cost of living,laborinputsin the
auctionsector,differentialpriceand incomeelasticitiesin the two sectors,
auction-pricedmaterialsthat are embodiedin finalproducts,variationsin
the outputof the auctionsector,changingrealcostsof producingcustomer
items,and inertiain wage increases.
39. With PcO= PAO = 1, the solution to the first-orderdifferenceequation for Pc is
log Pct = nit - m/s + (m/s)(l - s)t; while log PA, = mt.
Severalresultsappearto be robust for a wide class of such models:
(1) The rate of inflationcan be viewedas dependingon (a) the size of the
impacton the auctionsector,and (b) the speedof transmission(generally
throughwages) to the customersector. (2) A rise in the inflationrate
changesrelativeprices,becausethe auctionpricesget out in frontand the
othersdo not catchup. (3) Themorethe outputof a sectorcanrise,the less
its pricetendsto go up. (4) Realizedmarkupsare squeezedwherevercosts
are passedthroughwith a lag. (5) Inflationstartsslowly but builds up a
momentumthatkeepsit goingevenafterits initiatingsourceis eliminated.40
The impliedsqueezeon the profitmarginsof customerfirms duringin-
flationraisesinterestingquestions.To be sure, so long as their output is
risingrapidlyin an inflationaryboom, theirmarginsin the real worldare
buoyedup by a cyclicalproductivitybonus that does not appearin these
models.Yet the impliedinfluenceof inflationper se-as distinctfrom ex-
pandingoutput-in squeezingmarginsmay have some verisimilitude.It
clearlyappliesto public utilitiesand may apply to customer-strategy in-
dustriesthat are self-regulatoryin their passthroughof costs. And since
customer-strategy firmsaccountfor most U.S. equities,this squeezecould
even be one of the reasonswhy the stock marketdoes not performwell
duringinflationarytimes.4'
Becausecustomerpriceslag behind costs in these models, real wages
measuredin termsof those pricesare increasedby inflation,regardlessof
how wageratesthemselvesaredetermined.Thetypicalempiricalresultthat
real wagesdeflatedby total consumerpriceshave no distinctcyclicalpat-
tern may indeed suggestthat they are procyclicalin terms of customer
prices alone. Of course, this result conflictssharplywith the traditional
Keynesianview that employmentcould increaseonly with decliningreal
wages. But Keynes'belief in that classicalpropositionrested on the as-
sumptionthat firmstypicallyoperatedwith diminishingmarginalproduct
of labor,a thesisthat has long sincebeenexplodedby the facts on cyclical
productivity.41
40. These models closely resembleone constructedby James Duesenberryway back

in 1950. See "The Mechanicsof Inflation,"Reviewof Economicsand Statistics,vol. 32
(May 1950),pp. 144-49. Also, comparethe models used by RobertJ. Gordonto analyze
food inflationin "AlternativeResponses of Policy to ExternalSupplyShocks,"BPEA,
1:1975,pp. 185-94.
41. Note, however,that FIFO profitsarepulleddown only to the extentthat the cost-
passthroughlag is longer than the inventorylag. Severalaspects of cyclical movements
in profitmarginsare exploredby CharlesSchultzein his paperin this issue.
42. See Okun, "UpwardMobility,"pp. 211-13.
ArthurM. Okun 379
ASSETS AND INTEREST RATES
Wheninventoriesof auctionitems are broughtinto the analysis,they

play a key role, buildinga bridgeto monetarytheory. Supposeauction
marketsare "efficient"in the sensethatall worthwhileinformationis com-
piledand digestedby tradersandreflectedpromptlyin marketprices.Then
the demandfor inventoriesof auctionitems(like cotton)will be gearedto
priceexpectations;cotton(or a futurescontractin it) is potentiallya port-
folio assetfor anyonewho expectsits priceto soar.However,for customer
items(like dacron),whichare extremelyilliquid,stock demandsare essen-
tially transactionsbalancesof producers,distributors,and ultimatecon-
sumers;those stocks are likely to be much less sensitiveto inflationand
interestrates than are auction stocks. To be sure, cost expectations(for
wage rates and the like) can influencesellers,and price expectationscan
induceintertemporalshiftsby buyers(like the bulgein car salesin August
1974),but theirscope and magnitudeseem sharplylimited.
Given an emerginginflation,the responsesof asset marketsin a mixed
worldwill differfrom those of a pureauctionworld.In the latter(with a
sufficientsprinklingof assumptions),a jump in nominalinterestratesby
the expectedrate of inflationwill keep portfoliosin equilibriumfor every
assetexceptnon-interest-bearing cash.Theinducedeconomizingon money
holdingsproducesthe singleadverseallocativeeffectof inflationrecognized
in standardtheory-additionaltripsto the bank.Butthoseanswerscannot
applyto the worldof two classesof realassets,with the high inflationrate
on cottonandthelower(andlessrelevant)one on dacron.Nominalinterest
ratescannotriseenoughto equilibratethe demandfor cottonstocksat the
pre-inflationary levelwithoutcreatingan excesssupplyof loansanddacron
stocks.
As a result,the rise in nominalinterestratesmust leave some expected
net returnfromcotton speculationand thus maintainsome additionalde-
mandforcottoninventories,limitedby thehigh(andpresumablyincreased)
risk of holdingthese stocks. Sophisticated,wealthy,and risk-neutral(or
relativelyless risk-averse)investorsare handeda windfallby the oppor-
tunityfor speculationon cotton. Moreover,the buildupof cotton stocks
andanypossibleinducedholddown(or buildup)of dacronstockslengthen
thelist of realallocativeeffectsof inflation:therearechangesin the sched-
ule of tripsto lots of suppliers,not just to suppliersof money.
Whilethe expectedinflationrate on cotton thus sets an upperlimit on
the inflationpremiumin nominalinterestrates,I see no compellingreason
why the expected inflation rate on dacron should be a lower limit. Surely,
nothing guarantees or even suggests that the premium will correspond to
the weights of dacron and cotton in any overall price index, or that any
general "real" interest rate, however defined, will remain stable, or that the
premium thus will offer reasonable insurance to savers who want to hedge
against inflation. It is predictable that the structure of interest rates will be
altered, because the financial sector is also a mixed auction-customerworld.
Because thrift institutions borrow short and lend long, they must pursue a
customer "pricing" strategy, passing through the sluggish yields on their
portfolios in the form of sluggish interest rates to their depositors. The cus-
tomer strategy has more fundamental causes than the government ceiling
on interest, which is really an effort to preserve its viability. In any case, the
interest rate received by depositors does not compensate them at all fully
for increased inflation.
The role and importance of money and near-moneys become clear in a
mixed world. In a hypothetical world of universal, frictionless auction mar-
kets, every product is liquid, and a well-diversifiedportfolio of commodities
should dominate over demand deposits. It is precisely because customer
items are illiquid assets with sluggish prices that fixed-dollar assets are so
attractive to households who may need to spend for automobiles, appen-
dectomies, college tuition, life-insurance premiums, and property-tax bills
at uncertain times in the future.43
An inflationary world with greatervolatility of overall prices and of rela-
tive prices reduces the "real" services provided by liquid assets in ways that
do not get recorded on balance sheets. The typical household is offered no
reliable escape, even if it correctly anticipates inflation. Commodity mar-
kets are too risky; equity markets reflect the problems of customer-product
firms; borrowing more to accelerate purchases of customer-priced autos
and appliances is generally a losing strategy (1974 provided an exception);
only homebuying offers hope for salvation, and then only to a limited num-
ber of families in those intervals when mortgages are available. Ample
reasons emerge for the empirical fact that consumers save more in response
to previously unanticipated inflation.44 The diminished real wealth from
43. See Leijonhufvud,On KeynesianEconomics, pp. 80-81; Karl Brunnerand Allan
H. Meltzer, "The Uses of Money: Money in the Theory of an ExchangeEconomy,"
American Economic Review, vol. 61 (December 1971), pp. 784-805.
44. George Katona stressed that fact for years to a skeptical profession; see The
Poweifil Consumer(McGraw-Hill,1960).See also F. ThomasJusterand Paul Wachtel,
"Inflationand the Consumer,"BPEA, 1:1972, pp. 71-114.
ArthurM. Okun 381
nominal assets and the adverse "income effect" associated with greater risk
and uncertainty apparently swamp incentives to shift into goods. And
households may save particularly in liquid form in order to bolster their
dwindling ratio of cash reserves to income and consumption. Thus, in a
variety of ways, the new mixed world makes sense of the old story that in-
flation discriminates against the unsophisticated saver and investor, re-
distributing wealth from "suckers" to "sharpies."
Inflation may also stimulate investment in plant and equipment by auc-
tion industries and depress it in customer industries. The latter may well be
confronted with an increased real interest rate (in terms of their product
prices). Even if they are not, they may respond negatively to an equal rise in
nominal interest rates and inflation rates. A number of businessmen have
told me that their firms initially and tentatively evaluate a proposed new
plant by estimating how its output (at some "normal" operating rate)
would contribute to profits, assuming that the product will be sold, and all
inputs bought, at currentprices. That initial estimate of profitability will be
depressed when lenders are getting some inflation premium in nominal
interest rates-unless the firm assumes enough inflation of its product price
(relative to that of variable inputs) to recoup this inflation premium. As I
heard one businessman express his feelings, "I hate to make an investment
that can be bailed out only by inflation." In effect, even if his best estimate
is that he can push up prices fast enough to recoup the inflation premium,
he feels greateruncertainty about profitability and hence greater reluctance
to invest.
In summary, inflation has far more pervasive influences on portfolio
choices than the traditionally hypothesized shift from nominal assets to real
assets. It may push them away from nominal assets, customer items, equi-
ties of customer firms, and perhaps even physical capital of customer firms
into auction commodities and those few other real assets that are reason-
ably liquid, like real estate.
ADAPTATION AND ACCELERATION
The institutions of the customer sector are geared to a basically noninfla-

tionary world. And they have to adapt if the world becomes and remains
inflationary. Some of these adaptations are taking place today. Customer-
strategy firms are trying to widen their markups. They are shortening the
intervals at which they pass through cost increases; they change prices more
frequently; they are departing from historical cost calculations and trying
to anticipate future cost increases. Many firms have begun to think and
price LIFO as well as to report LIFO profits; they are focusing anew on the
gap between replacement and original costs of capital in setting prices;
some have ended the practice of accepting orders with fixed delivery prices,
while others are putting future cost increases into those prices; some firms
have instituted wage and pension escalators and have shortened labor con-
tract periods. They are searching for new formulas and yardsticks to sup-
plant the dollar yardstick. More and more investors are learning to partici-
pate in commodity markets and real-estate ventures. Lenders who need
certainty of capital value are increasingly reluctant to make long-term
commitments.
Most of these adaptations shorten the lags and increase the intensity of
inflation associated with any given level of real economic activity. Thus,
they generate an adverse shift of the short-run Phillips curve that reflects
the inflationary experience of the past. That shift accords with the predic-
tion of the accelerationist model, although the mechanism of acceleration
is very different from that implied by a natural unemployment rate and a
particular expected inflation rate. I believe the United States is currently
suffering from an acceleration resulting from such adaptations to past in-
flation, as well as from the momentum of the more normal long lags in
customer and career markets.
The gradualness and the costliness of these adaptations reveal the great
dependence of customer institutions on the dollar yardstick. The adapta-
tions require new techniques of management control, new-and perhaps
inherently less efficient-kinds of customer and employee relationships,
and a new breed of accountants. The innovators find themselves out of step
and unable to communicate with their customers, suppliers, lenders, and
government tax collectors or regulators. And so the economy adapts to
inflationat only a tortoise's pace, even though everyone expects inflation to
continue at least at a mule's (if not a hare's) pace.
Implicationsfor Welfare and Policy
Customermarkets are valuable institutions to society, because informa-

tion costs are really high. In comparison with a dominantly auction econ-
omy, they cut the cost of shopping, and reduce the resources devoted to
Arthur M. Okun 383
trading in spot and futures transactions and to the negotiation of formal

contracts. They lower job turnover and create reasonably predictablecareer
income profiles; they encourage firms to invest in workers and to build
career ladders for them. They economize on resources devoted to point-
estimate forecasting of the future. They promote reliance on custom and
habit. They allow accounting systems and hence orders, capital budgeting,
and financing to be guided by reliable historical costs rather than conjec-
tural future costs. They make money and, more generally, fixed-dollar as-
sets valuable and reasonably reliable claims over a wide range of consumer
goods and services that may be desired many years in the future.
Prolonged and intense inflation upsets many habits of economic life, con-
fronting consumers with price increases and price dispersions that send
them shopping; making them doubt their ability to maintain their living
standards, and downgrade the value of their career jobs and long-term
savings; and forcing them to compile more information and to try to pre-
dict the future-costly and risky activities that they are poorly qualified to
execute and bound to view with anxiety. The recognition by the consumer
that economic institutions are gravely disturbed by inflation is an apprecia-
tion of reality-not money illusion. The illusion-Walrasian-general-
equilibrium illusion or barter illusion-lies in the models of an economy
in which inflation does not matter, offering automatic protection to savers
through the interest premium on nominal assets and leaving intact the
relative prices of cotton and dacron and the relative wages of janitors and
professors.
The accelerationist is right that, in some respects, people are fooled by
inflation. But the fooling lies, not so much in disappointing predictions of
precise outcomes as in undermining the foundations of habit and custom
that permit one to live without many point-estimate predictions. And the
institutions of the system provide no vehicles of "unfooling" to transport
the economy to a no-tradeoff situation of "fully anticipated" inflation and a
natural unemployment rate.
I do not know, at this point, how to quantify the benefits of a customer
economy or how to assess the fraction of those benefits that would be dissi-
pated by the persistence of rapid and erratic inflation. Clearly, the total
benefits are large: consider the welfare loss that would be imposed on a
person forbidden ever again to engage in any economic transaction with
anyone whom he has voluntarily bought from or sold to in the past-his
employer, banker, mechanic, tailor, and the like. The percentage "discount"
imposedby intenseinflationon these relationshipsis not enormous,but

neitheris it trivial, given their dependenceon the dollar yardstick.As
WilliamBrainardsuggestedin conversation,estimatesof the substantial
costsassociatedwitha shiftto the metricsystemmaygive somecluesto the
muchlargercosts of abandoningthe dollaryardstick.
Howeverthese welfarebenefitsare evaluated,they must be balanced
againstthe welfarecosts of a customereconomy-greatervolatilityof out-
put andemployment.In a recessionaryworldof universalauctionmarkets,
price and wage levels would plunge rapidlytoward zero, allowing the
Keyneseffectandthe Pigoueffectto restorefullemployment.45 And rapidly
risingpricescould similarlyhalt booms with a brief intensespurt(rather
than a long naggingbout) of inflation.
AlthoughI cannotprovethat the prevalenceof customermarketsyields
a net benefit,subjectivelyI thinkthe systemis worthsaving.The abilityof
the systemto limit the waste of idle resourcesand to contain the infla-
tionarybias in wageslooked prettygood in the fiftiesand sixties,particu-
larlyin contrastto the recordof the thirtiesand the forties.Indeed,in that
tranquilperiod,long-runattachmentsmust have paid off well. Infrequent
layoffs,low quitrates,good internalladdersservedfirmsand careerwork-
ers, while sluggish cost-basedprices satisfiedcustomersand suppliers.
Those successesmay well have graduallyencouragedmore customerand
careerstrategies;andthat,in turn,maycontributeto the findingthatwages
andpriceshavebeenless responsivein recentyearsto givencyclicalmove-
ments of real activity.46
STEADY INFLATION
Someproposedpolicystrategieswithrespectto inflationseekan alterna-

tive yardstickfor the dollar. The proponentsof steady inflation-rather
than low inflation-would set a targetof stabilityin the firstderivativeof
the valueof the dollarratherthanin its level. The steadilyshrinkingdollar
45. To do so, they would have to outweigh any negativeimpact of deflationaryex-

pectationson investmentdemand and have to escape any liquiditytrap. Like Robert
Hall (in his articlein this issue), I would expect a high degreeof downwardwage and
priceflexibilityto restorefull employment.But, unlikehim, I would attacha largesocial
cost to wage and price volatility.
46. See Cagan,Hydra-HeadedMonster,pp. 2-8. Tobin pointsto the decliningagricul-
turalshareof employmentand GNP in the postwarera as anotherpossiblecontributor.
See his "Inflationand Unemployment,"p. 14.
ArthurM. Okun 385
maybe a reasonableyardstick,but it has to be secondbest-just as a 5 per-
cent annualdeclinein the lengthof a "yard"wouldbe inferiorto stability.
I can see two possibleargumentsfor the compromisegoal of steadyin-
flation,but I findneitherpersuasive.First,if the key difficultyof stabiliza-
tion stemmedfromthe downwardrigidityof the level of wages,it could be
sensibleto aim at a normalinflationrate that kept nearlyall wage move-
mentsin the positiverange-so that absolutecuts in wage ratesin signifi-
cantsectorswouldnot be neededto achievethe desiredmacro-equilibrium.
Second,if the economywerefairlywelladaptedto someparticularinflation
rate(like5 percent),it mightnot be worththe transitionalcost to wringthat
inflationout of the system.As I have arguedabove, however,the adapta-
tions arelimitedadjustmentsto varyingand uncertaininflation(whichstill
leave the systemheavilydependenton the doliar yardstick),ratherthan
full adjustmentsto any particularinflationrate.
The main problemof steadyinflationas a goal is its lack of credibility.
Targetingon a stable first derivativeis admittingfailurein the effort to
stabilizethe level. Why should anyone expect any greatersuccessin sta-
bilizingthe rate of changeof the price level than in stabilizingthe price
level?Moreover,in the contextof the model of this paper,steadyinflation
requiressuch a full adaptationthat the same expectedinflationrate can
pertainto dacronand cotton, and to the wagesof professorsandjanitors.
Steadyinflationstrikesme as even more of a miragethan when I first
calledit that four yearsago.47
INDEXATION
Another proposedstrategyfor policy would substitutefor the dollar

yardsticka new cost-of-livingunit throughindexationof incomes,assets,
debts, and taxes. Thus far, explicitcost-of-livingadjustmentshave been
enactedfor social securityand for federalretirementbenefits,and both of
the formulaswrittenintolawinvolveserioustechnicaldefects.48Apparently,
constructionof a substituteyardstickis not a trivialtask.
The most far-reachingproposalwould promotewage indexationas a
generalpractice.Throughindexation,careerwages would respondmore
47. See "Mirageof Steady Inflation."
48. The defect in social security is discussed in Barry M. Blechman, Edward M.
Gramlich, and Robert W. Hartman, Setting National Priorities: The 1976 Budget
(Brookings Institution, 1975), pp. 175, 182; that in federalretirementintroducesa re-
currentand compoundedroundingerror.
promptlyandmorefullyto changesin the pricesof auctionproductsandin

flexiblewages.As RobertJ. Gordonhas convincinglyargued,thatmustbe
sociallyundesirablewhenthe changesin auctionpricesresultfromchanges
in supply.49But when the price changes are demand induced, presumably
indexedwageswouldmorenearlyapproximatethe "shadowequilibrium"
wagesof an auctionlabormarket.
A programof universalwage escalatorswould representan effort to
breathelife into the textbookauctionmodel (accompaniedby a hope that
supplyshifts would rarelybe the culprit).In the indexedworld, like the
auctionworld,the pricelevel wouldbe volatileand wouldrespondrapidly
to aggregatedemand;hence,a nonaccommodating monetarypolicywould
automaticallyunleashthe Keynesand Pigou effectsin the earlystages of
inflation.Moreover,becausethe decreasedinflationresultingfromrestric-
tive discretionary policieswouldemergewitha muchshorterlag behindthe
reducedoutput and employment,such policies might becomepolitically
more feasible.The thesisis that, if inflation"hangsout" sooner,it will be
stoppedsooner.I find the argumentintriguing,but not persuasive.
The equitycase for wage indexationseems even less persuasive.To be
sure, escalatorsmake the real wage of an indexedworkerless dependent
upon sucharbitraryelementsas the timingof his contractor the effective-
ness of his unionleader.And they may help to resolveconflictby keeping
inflationissuesout of the wagebargain.At the sametime,by increasingthe
volatilityof price movements,they must exacerbatethe uncertaintyand
variabilityof realincomesfor those who have no escalators.If indexingis
less than universal,its equityeffectsmust be uncertain.
And if indexingbecameuniversalfor wages,contractsfor futuredelivery,
claims,and taxes,the market-basketunit wouldbecomethe yardstickand
the numeraireof the economy,and the price level would becomehighly
unstable.Clearly,the pricesystemwouldgo haywireif all the sellersof the
itemsin a comprehensivemarketbasketset "realprices"in market-basket
units(only by magicwouldthe pricesadd up to unity).Now supposethey
aimcollectivelyat a realpricethat exceedsunity,but theyhaveto set price
tags in dollars.If everybody'spriceis gearedto full escalationwith a short
lag, the inconsistencyof real aims must resultin explosiveinflation.The
pricesystemcan workwithgold, peanuts,or dollarsas the standard,but it
cannotworkon an all-inclusivemarket-basket standard.To put it another
49. "AlternativeResponses of Policy," pp. 201-02.

ArthurM. Okun 387
way, the moneyequationwould be pushedout of the general-equilibrium
system,makingthat systeminsoluble.
All thingsconsidered,I suspectas good (or as bad) a case can be made
for a socialeffortto discouragethe responsivenessof careerwagesto cost-
of-livingmovementsas for one to promoteit. If the link could be broken
entirely,inflationwouldbe morenarrowlyconfinedto auctionmarketsand
limitedby a differentkindof automaticstabilizer-a squeezeon consumer
demand due to depressed real labor income. Transitoryinflationary
shocksmightthen blow themselvesout withmuchless impacton the price
level.Thepoliticalprocessmightevenbe improvedbecauseinflationwould
squeezethe majorityof votersharder.
POLICIES FOR THE AUCTION SECTOR
If instead of designingalternativeyardsticks,policymakersshould be
aimingto save the dollaryardstick,then they must recognizethe special
role of the auctionsector.While auctionitems may not be the sourceof
excessdemand,theywillbe the sourceof majorpriceincreases.Speculation
on auction-productinventoriesin a strong cyclicalupswingis not a re-
sponse to any genuinechangeof relativescarcities,and servesno social
functionthatI cansee.Hugepriceincreasesof auctionitemsimposemacro-
economicwelfarecostsby transmittingboth inflationand,throughrestric-
tive fiscal-monetary to customerand careermar-
policies,underutilization
kets.50 For thesereasons,publicstocksof auctionitems,measuresto ensure
adequateand reliabledomesticsupplies,and governmentdisincentivesto
cyclicallyspeculativestockbuildingall belongin the kit of economicsta-
bilizationinstruments.
Theserealitiesof the mixedauction-customer worldhavebeenflagrantly
ignoredin U.S. agriculturalpoliciesof the seventies.Indeed,it is hardto
find mistakesin monetaryor fiscal policies that can match the macro-
economicdamagewreakedby the policiesof exportpromotion(backedby
no publicstocks) of farmproducts.Certainpolicies on the pricingof oil
50. The macroeconomicexternalitiescan be illustratedby the following arithmetic
example. Suppose (1) the price elasticity of U.S. demand for some auction item (say,
wheat) is 0.2; (2) in light of the social attitude toward the inflation-unemployment
tradeoff, stabilization policy aims at a given nominal GNP target. Then the production
and domesticsale of one extra bushel, pricedat $4, lowers nominal farm GNP by $16
and permitsa $16 increase in nominal nonfarm GNP. So the bushel is really worth
about $20!
that essentiallyignoredtheirmacroeconomicinflationaryand recessionary

effectsrate anotherhigh place on the horrorlist.
Theespeciallyheavyweightof auctionitemsin U.S. exportsandimports
hasimportantimplicationsforexchange-rate policiesandforthemonetary-
fiscalmix. The discouragementto importsfrom a depreciatingcurrency
may sacrificepotentialanti-inflationary benefits.Thus,the undervaluation
of the dollaron tradeaccountin 1974and 1975may be significantlyim-
pairingU.S. priceperformance(giventhe level of economicactivity).
The anti-inflationarybenefitsof a highly valued currencyare also an
argument-although not necessarilya decisive one-for less monetary
stimulus(offsetby morefiscalstimulus).Withflexibleexchangerates,high
interestratesthat encourageshort-termcapitalinflowsmay strengthenthe
currencyand help particularlyto moderateinflation.Such a shift in the
monetary-fiscal mix wouldalso help if domesticinventoryholdingsof auc-
tion commoditiesareespeciallysensitiveto interestrates.In general,policy-
makersin the mixedcustomer-auction worldshouldbe criticallyawareof
the possibilitiesfor shiftingthe inflation-output
tradeoffwithmeasuresthat
have their impactprimarilyon either auction-pricedor customer-priced
items. Many of those options lie in internationaleconomicpolicies, and
some wouldrequirenegotiationand coordinationamongtradingpartners
to avoid "inflate-thy-neighbor" strategies.
As has beenwidelysuggested,the worldwidecharacterof the 1973boom
and the 1971-73 devaluationsof the dollar contributedto the particular
severityof recentinflationin rawmaterialsand othertradedcommodities.5'
The story of 1966may have beenjust the reverse:excess demandin the
UnitedStateswas not sharedby its tradingpartners,andthe dollarwas in-
creasinglyovervalued;as a result,importspouredin, dampeningdownthe
inflationrate on commodities.Indeed,commodityinflationwas unusually
moderatein 1966-69;as shownby figure1, it did not even matchthat of
the middleand late fifties.That stabilityin the auctionsector may have
contributedto the initialcomplacencyof U.S. policy.Thatexperiencealso
contributedin 1972-73 to the seriouserrorsof economicforecasters(in-
cludingme). We wereblindsidedby commodityinflationwhilewe focused
happilyon tranquillabormarkets,as thoughthey werethe only potential
sourceof seriousinflation.
51. See, for example,WilliamNordhausand John Shoven, "Inflation1973:The Year

of Infamy,"Chlallenge,vol. 17 (May/June 1974), pp. 18-19.
ArthurM. Okun 389
FISCAL POLICY
Themakersof stabilizationpolicyfacea particularproblembecausecus-

tomerpricesand careerwagesare laggingindicators.By the time they ex-
hibit a visibleaccelerationin a cyclicalexpansion,inflationwill have be-
comedeeplyentrenched.Discretionarypolicydecisionsrequirethe guideof
moresensitiveindicatorsthat will registerexcessdemandmorepromptly;
quit rates, orderbacklogs,and the like should be examinedcarefullyfor
theirpotentialcontributionto an earlydiagnosis.
The reverseproblemcreatedby customerand careermarketsis the need
for so much unemploymentfor so long in orderto work off an inflation
that has become well entrenched.Under those circumstances,there are
overwhelmingmeritsin fiscal devicesthat would operatedirectlyto hold
down priceswith less compressionof output and employment.The ideal
medicinefor an economyin whichboth unemploymentand inflationare
too high is a value-addedsubsidy.And the nextbest is reductionin broad-
basedexciseandsalestaxesandpayrolltaxes.Whiletheywouldbe harmful
in excess-demandinflation,these are the only prescriptionsthat deal di-
rectlyandefficientlywiththe diseaseof stagflation.Theone objectionI hear
again and again in responseto my advocacyof such measuresconcerns
theirnovelty-a disadvantagethat wouldbe remediedby theiradoption.
INCOMES POLICY
Finally,a casefor price-wageinterventionby the governmentemergesin

the customer-auction world.If most wagesare heavilyinfluencedby some
criterionof fairness(and do not registersome preciseoptimizinginter-
sectionof supplyanddemand),the governmentcanbe constructiveby pro-
motinga noninflationary criterion(or "guidepost").In a worldwhere,for
good reasons,peoplecareaboutthe stabilityof the pricelevel, everyprice
or wage that has a zone of indeterminacyimposes an externalmacro-
economiccost whenit is set at the top of the zone andyieldsa socialbenefit
whenit is set at the bottom.To be sure,thereare gravedifficultiesin pub-
licly influencingtheseprivatedecisions;and misconceivedcontrolscan do
even moreharmto customerand careerrelationshipsthan inflationdoes.
But the potentialsocial surplusfrom recognizingthe macroeconomicex-
ternalitiesmay be large enough to compensatepartiesthat accept some

restraints.62
The worldof mixedcustomerand auctionmarketsposes seriousdilem-

mas for publicpolicy.It is not the worldof costless,fullyadjustedinflation
or of intractablenatural unemploymentrates. Its tradeoff problem is
genuineand genuinelyagonizing.The systemneeds reasonableprice sta-
bility to preserveits institutionalframework:it cannot thrive with a 14
percentor even a 7 percentrate of inflation.And some loss of outputand
employmentis worthwhileand unavoidablein the cure of an entrenched
inflation.Butbecauseso manypricesandwagesrespondso littleto demand
in the shortrun,holdingdownoutputandemploymentis an extremelyslow
andexcruciatingly painfulcurefor inflation.Theprolongedmaintenanceof
unemploymentratesof 7 percentor more,and of shortfallsin annualout-
put of $100 billion to $200 billion, is neithera sensiblenor a credible
strategy.Fortunately,the systemoffersmany opportunitiesfor more effi-
cient policiesto dislodgean entrenchedinflation.These includemeasures
that counterthe destabilizinginfluenceof auctioncommodities;that use
taxes and subsidiesto reducecosts withoutreducingaggregatedemand;
and that influencewageand pricedecisionsto reflecttheirmacroeconomic
externalities.
52. See my 1974proposalfor realwage insurancein ArthurOkun,"IncomesInflation
and the Policy Alternatives,"in "The Economists'Conferenceon Inflation: Report,"
vol. 1 (1974; processed),pp. 369-71.
Comments
andDiscussion
WilliamFellner:ArthurOkun'spresentcontributionmay be viewedas a
companionpieceof Hicks'workon similarproblems,thoughwitha num-
berof distinctivefeatures.My firstcommentwill relateto the paper'smain
analyticalthesis;my second and third commentswill be concernedwith
alternativepolicy approachesto the inflationproblem,in view of Okun's
observationson these.
First,Okun'ssellerin a customermarketis uncertainaboutthe long-run
price elasticityof the demandthat he individuallyfaces, and he is risk
averse.The reasonsOkungives for the uncertaintyabout elasticity-even
about its sign-are presentedin a particularlyconstructiveand original
sectionof the paperin whichthe disruptiveeffectof pricechangeson cus-
tomerrelationsis stressed.
As for the seller'sriskaversion,this is in partthe kindthat is consistent
withthe von Neumann-Morgenstern-Savage axioms,butthatriskaversion
also includesanotherelement.Even if the risk-averseseller acts consis-
tentlywith the usual probability-utility axioms,he will prefersmallerex-
pectedprofitspredictedwith relativelylittle uncertaintyto largerexpected
profits with undesirablecharacteristicsof the highermomentsof a dis-
tribution.
But Okun'sstrongemphasison custom,and on past practicesthat have
cometo be consideredusualand fair,stronglysuggeststo me that,accord-
ingto him,moreis involvedherethanthe riskaversioncompatiblewiththe
usualprobability-utility axioms.One arrivesat the sameconclusionwhen
readingHicks. They have their way of describingthe additionalfactor
shapingthisriskaversion.Letme saya wordaboutmy wayof lookingat it.
Basically,the questionhereis how to interpretthe conceptof rationality.
Onemustrecognizethatit is oftentoo costly,or evenliterallyimpossible,
391
to structurea probleminvolvingprobabilitiesin such a way as to derive

probabilityjudgmentsfor the payoffs from the probabilitiesassignedto
clearlyspecifiedconditions(andconditionsof conditions),and thento add
togetherthe weightedconditionalprobabilitiesso defined.Many decision
problemsremainill-structuredin the sense that one has little confidence
that all conditionsto whichprobabilitiesshouldhave been assignedhave
indeedbeen accountedfor, and this state of affairsdoes not lend itself to
even roughsubjectiveestimatesof the variancethat is therebyintroduced
into the final probabilityjudgment.Also, the final probabilityjudgment
relatingto the payoffsis apt to be more controversialamong competent
individualsthe less well-structured a decisionproblemis in this sense.
In some of thesesituationsit is verytemptingto rely on a rule of thumb
biasedtowardthe statusquo,in somesenseof thisterm,andI see no reason
for postulatingthat such behavioris generally"unintelligent"or tends
toward worse long-run results than reliance on controversialpersonal
probabilities-on hunches.This poses a problemof considerableinterest
for decisiontheory,andalso for the theoryof resourceallocationand other
branchesof economictheory.To my wayof thinking,Okunis callingatten-
tion to a specificaspectof this generalproblem.A quantitativeappraisalof
the problem'ssignificancefor pricingpracticeswill requiremoreempirical
research;the models sketchedin Okun'spaperare intendedto be merely
illustrative,and they performthis functioneffectively.Inflationmakes it
impossibleto follow variouscustomarypracticesthat have beenpreferred
on both sidesof a good manymarkets,thoughthe evaluationof theseprac-
tices froma generaleconomicpoint of view raisesas yet unresolvedques-
tions.
Second,turningto the inflationproblemmoregenerally,I will note that
both Okunand Hicksare removingfromthe Keynesiansystemthejustifi-
cationfor inflationthat is incorporatedin Keynes'conceptof involuntary
unemployment (adaptedto the growthcontextby Tobin).Therethejustifi-
cationis providedby postulatingthatrisingpricesbringreal wageratesto
their equilibriumlevel. Okun'sanalysissuggeststo me that he is rightly
skepticalabout this hypothesis-certainly about building policy on it.
Hicks,too, expressedhis doubtsaboutthe real-wageequilibratingeffectof
inflation.
I don'tsee anyotherconvincingcasefor net benefitsfroma risinggeneral
pricelevel.The remainingargumentsseemto me clearlytoo weakto offset
the generallysharedgravemisgivingsabout the destabilizingeffectsof in-
ArthurM. Okun 393
flation.Onemustthereforeconcludethat fromabout 1965on inflationhas
resultedeither from the mistakenunwillingnessof demand-management
policymakersto observeconstraintsset by price-levelobjectives,or by the
inability of a policyaimingfor pricestabilityto achieveit and to avoidthe
unintendedresultof underutilization of capacitycombinedwith inflation.
Thedatafor 1951-65discloseneithera mistakenunwillingnessto be guided
by price-stabilityconstraintsnor lack of successin this effort.The more
recentrecordis different.
Given the 1965-74record,the problem,as I see it, is to conditionthe
marketsanewto price-levelobjectivesof policymakers,as the only way to
avoid disturbancesfrom large unexpectedmovementsof prices. In my
appraisal,a gradualreturnto pricestabilityis the appropriatetarget,and
Okun'sanalysisstrengthensthis belief. Even the dissentersshould agree
thatdemandpoliciescanbe successfulonlyif theyhavesomecredibletarget
for prices.Thereasonis thatpriceexpectationsmustbecomegroundedat a
point lying outsidethe area of the expectationalsystemitself.
Third,a demand-management policyto whichpriceexpectationsbecome
conditionedcan be employmentorientedonly withinthe limitscompatible
with a priceconstraint.Such a policy cannotpromisethe achievementof
specifiedunemployment-rate targets.The policy wouldhave to be supple-
mentedby workablearrangementsinvolvingde facto subsistenceguaran-
tees,and,to be successful,arrangementsof this sort needto satisfya num-
ber of specificconditions.But this is the best chancewe have, becausethe
attemptto hit low unemploymentratesby meansof inflationis doomedto
failureand directpriceand wagecontrolswouldnot solve our difficulties.
By now the trulycontroversialquestionrelatesless to the dangersof infla-
tion thanto attitudesaboutcontrols.This,too, is reflectedin Okun'spaper.
Okunis certainlyno believerin comprehensivecontrolprograms,but I
havestrongmisgivingsalso aboutbenignmethodsof administering limited
wageand pricecontrols,includingthe acceptanceof governmentalrespon-
sibilityfor specificwage-pricedecisionsin the form of guidepostpolicy,
jawboning,and the like. Okundoes not sharethesemisgivings.In my ap-
praisal,directcontrolscan achievecertainobjectivesonly by relianceon
the police power-and then only at the expenseof otherimportantobjec-
tives. Tryingto be "nice" about the administrationof such a program
makesthe regulationslargelyineffectiveand theirincidenceentirelyhap-
hazard.The main resultof this ineffectivenessand of the hit-or-missexer-
cise of governmentalpowersis demoralizationin a no-man'sland. Judging
by the policyconclusionsOkunderivesfromhis analysis,whathe andI see
differentlyis the magnitudeof this risk, whichto me seemsvery great.
MichaelWachter:Okun'spaper is very much in the spirit of his earlier

paper,"UpwardMobilityin a High-Pressure Economy"(BPEA, 1:1973),
and representsan importantextensionof the ideas exploredthere. Here
Okunaddressestwo issues-the mechanicsand the welfarecosts of infla-
tion-and applieshis model to see its implicationsfor these problems.I
thinkthat the papermakesconsiderableprogresson both issues,and pro-
videsa veryusefulframeworkfor understanding the institutionaleconomic
environmentin whichinflationtakesplace.In particular,it is a neoclassical
modelthatdoesnot relyon thejob-searchimagery.My disagreements with
thepaperaregenerallyconfinedto a few of the inferencesthatOkundraws
fromhis model. In these cases Okunstraysfrom what I believeto be the
(deceptively)strongneoclassicalconclusionsinherentin his model.
The model is based on two sectors:a customer-market sector and an
auction-marketsector. A particularlystrong point is Okun'ssystematic
descriptionof the natureof these polar markettypes and theirlikely im-
pacton inflation.First,the existenceof customermarketscreateslong lags
betweeninflationand its determinants.AlthoughOkunuses the terminol-
ogy "costsof information"to referto the basis of the laggedresponse,he
differentiateshis approachfromthejob-searchschool.I wouldargue,how-
ever,that this terminologycan be seriouslymisleading.In the job-search
school, "costsof information"refersto the costs workersand employers
(buyersand sellers)incurlargelyin learningaboutthe currentdistribution
of prices,wages,employmentopportunities,and so on. Longlags aremost
unlikelyto arise in this situation;indeed, these informationalproblems
may characterizethe rapidly respondingauction marketsof the Okun
model.The lags in the customermarketsarisein an institutionalsetting.
Buyersand sellersdo have informationalproblemsin understanding their
currentenvironment,butthe realissueis thattheyaremaximizinglong-run
profits(or wagesand so on) in a worldof ongoingrelationshipsamongthe
economicactors. Factors such as customerloyalty, specifictrainingof
workersand theirupgradingalongpromotionladders,and the prevalence
of contractualobligations(explicitandimplicit)suggestthatthe realprob-
lem is thatthe future-ratherthan the present-is unknown.The absence
of futuresmarketsthat would enable economicagents to hedge against
mostchangesin pricesandwagesandin performancelevelsof workersand
firmsis not accidental.Theoretically,of course, it could all be done by
ArthurM. Okun 395
futuresmarkets;empirically,it cannot-the possible futurestates of the
world are too numerousand complicated.In this sense, it is the institu-
tionalarrangements of the customermarketsand not costs of information
thataccountfor the sluggishresponseof inflation(as wellas anynumberof
other variables).My disagreementwith Okun here is not merelytermi-
nological,but it is only a matterof degree:I wouldplace more weighton
the inabilityto forecastor to hedge against future developmentsin an
economyin whichongoingrelationshipsare prevalent.
Okun'ssecondmajorpoint is that inflationhas large welfarecosts be-
cause it impingeson the workingsof these long-runrelationshipsin the
customermarkets.In this argument,I believethat he is directlyon target.
I wonderwhetherthis newpaperimpliesa shiftin Okun'sown utility-func-
tion tradeoffbetweeninflationand unemployment.In the world of cus-
tomer marketsthere is no such thing as a pure inflation.Institutional
arrangements can be alteredto meetnew problems,but the transitioncan
be exceedinglylong and difficult.Furthermore,no evidencesuggeststhat
thereare "inflation-neutral" methodsof institutionalchange.And even if
they exist, it is impossibleto predictthat they will be adopted,given the
lack of a theory of institutionalchange.Of course, the quantitativeim-
portanceof this problemis not easilymeasured,so that its impacton the
neutrality-of-inflation storycannotbe assessed.
An importantand valuableaspect of Okun'swork has been to divert
concernwith the problemof institutionallags away from oligopoliesand
unions and toward broaderconceptssuch as customermarkets.In this
mannerthe model focuses on the desireof firmsto deal with promotion
ladders,quits, and turnovercosts, ratherthan union contracts.This con-
structsuffersfromthe uncertaindefinitionof its boundaries,andthereis a
relatedunresolvedproblemof endogeneity.Whereasthe unionand oligop-
oly sectorsareidentifiableandthereareevensometheoriesaboutwhythey
arise,the samecannotbe said(at leastto the sameextent)for the customer-
marketsectors.For example,what industriesdevelopinto customermar-
kets and why? And what characteristicsare customermarketslikely to
have-which partybearsthe risksof rigidpricearrangements, the costs of
trainingworkers, and the like? This is an importantgap in the analysis,
althoughone certainlycannotfaultOkunfor not solvingit all in one paper.
In Okun'smodel,customermarketsare not spreadrandomlyacrossthe
economy.Rather,the complicatedarrangements of thesemarketsdevelop
as an (internalmarket)efficiencyresponseto whatwouldbe externalitiesin
the auctionmarket.'Clearly,microeconomicsand profit-maximizing be-

haviorare relevantin this world.I am reluctantto arguethat all institu-
tional arrangements,along the customer-auctioncontinuum,are in equi-
libriumin the third quarterof 1975; but the role of efficiencymakesme
cautious about suggestingchanges in these arrangementsuntil I know
wherethe inefficienciesare located.It is on this point, for example,that I
would criticizeOkun's suggestionson the usefulnessof wage and price
controls.He arguesthatwagesareheavilyinfluenced,andpricessomewhat
influenced,by some criterionof "fairness"in customermarkets.Un-
doubtedly,he is correct,but one must exercisecare in interpretingwhat
fairnessmeans.
"Fair"wage(andprice)differentialsdo not implydifferentialsthat arise
largelyby chance,play a smallallocativerole, or respondeasilyto govern-
mentmanipulation.Thisproblemis madeall the moredifficultby the cycli-
cal variationsof thesedifferentialsas a consequenceof the differentlengths
of the planningperiodof the relevantfirms.As a result,the government
cannot observeand then enforceequilibriumdifferentials.Beyondthis is
the enforcementproblemposedby the sheernumberof wagesandpricesto
be supervised.Hence, I interpretthe Okun customer-auctionmodel as
making an incomes policy less ratherthan more attractive.Wages can
alwaysbe controlledby the government,but the socialand economiccosts
may be high in the customermarkets.
In any case, to the extentthat controlsare adopted,they need not be
economy-wide.As I have arguedelsewhere,an analysisof wage differen-
tials makesit clearthat cost-pushinflation,to the extentthat it currently
exists,is heavilyconcentratedin theconstructionandgovernmental sectors.2
Thewagedifferentials in thesetwo sectorshaveshowna secularrisethatfar
outmatcheschangesin othersectorsoverrecentyears.The wageproblems
in these sectors,however,do not arise specificallybecausethey are cus-
tomermarkets.Rather,inappropriategovernmentpolicy seemsto be the
sourceof the inflationarybias. Consequently,overalleconomy-widecon-
trolsarenot needed;instead,structuralchangesin thesetwo marketscould
go far in reducingcost-pushfactors.Concentratingcontroleffortsin the
offendingsectorsalsohas themeritof easingthe enforcementproblem.Un-
fortunately,such a simplesolutionseemsto be politicallyunattractive.
1. The efficiencyrole of internallabor marketsis a primaryargumentin my paper,

"Primaryand SecondaryLabor Markets: A Critiqueof the Dual Approach,"BPEA,
3:1974, pp. 637-80.
2. This argumentis documentedfurtherby Robert Hall's paper in this issue.
ArthurM. Okun 397
I wouldalso arguewith Okun'sconjecturethat customermarketshave
an inflationarybias. It may very well be that thereare significantinstitu-
tional rigiditiesto fallingwage levels (but muchless so to prices),so that
customermarketsmay be proneto wageinflationat verylow ratesof infla-
tion. On the otherhand,I see no reasonwhycustomermarketsshouldim-
part a significantupwarddrift to inflationratesthat are not close to the
zero level. In particular,the U.S. economyhas had little experiencewith
wagedrift(or overfulfillment of contracts),andincreasingunion-wagepre-
miumshavebeenlimitedlargelyto the constructionandmunicipalsectors.
I believethat the inflationproblemof customermarketsis more closely
relatedto the timingof discretewage and pricechangesthan to a secular
upwardbias.My guessis thatthe currentpriceincreasesin steelandalumi-
num are examplesof just this point.
I foundthe sectionon assetsand interestratestoo brieffor such a diffi-
cult question.This topic alone could be the subjectof a separatepaper.In
any case, I do not follow Okun'squickjump to the conclusionthat the
mixed world of customerand auction marketsdiscriminatesagainstthe
"typical"household-that is, favors the "sharpies"over the "suckers."
Empirically,the literatureon the redistributionaleffectsof inflationargues
stronglyagainst a systematicbias in favor of the wealthy.Okun's dis-
cussion of the reactionof businessmento inflationis more in the spirit
of his argument.Institutionalrigiditiesmakeit difficultfor firmsto adjust
to a changein the rate of inflationeven whenthe new rate is equalto the
ratethey estimated.Hence,in a mixedworldof rapidlyand slowlyadjust-
ing markets,inflationgeneratesconsiderablerealeffects.Theseeffectscan-
not be anticipated,however,in largepartbecausetheydependuponunder-
lying secularchangesin relativeprices(or productdemands)that may be
takingplaceas wellas becausenot all customermarketsadjustslowlyall of
the time. Whencontractscome due or whennew commitmentsare made,
the relevantcustomermarketsadjust rapidly.The result-and this is a
centralpoint-is that the real effectsof inflationgeneratea greatdeal of
uncertainty.
A few smallpoints: First, Okundoubtsthat increasingtransfersto the
unemployedcontributeto an increasein the noninflationaryunemploy-
mentrate,buthe subsequentlyarguesthatlayoffsfromthe customersector
are not as painfulbecauseof transferincome.Althoughthis is not neces-
sarilya contradiction,I wouldarguethat the (relative)increasein transfer
incomeis likely to have the formeras well as the lattereffect.
398 BrookingsPaperson EconomicActivity;2:1975
Second,Okunarguesthat wagesessentiallyfollowsomecombinationof
pricesand otherwages,but are largelyindependentof aggregatedemand.
Indeed,even the reduced-formrelationshipbetweenpricesand past and
currentwages is unstableover time and across countries.Althoughthe
underlyingmechanismis quite different,by adoptingthis approachOkun
is utilizingthe Keynes-Weintraub type of model,in whichwagesareexoge-
nous.This strongexogeneityargumentis, in a sense,a natural-ratetheory
of the rate of wage inflation.It is even more pessimisticthan the Lucas-
Sargentmodels in that they hold out the hope that at least the reduced-
form, autoregressivestructurewill be usefulfor predictionpurposes.In a
paperon the responsivenessof wagesto inflationfor the next Brookings
panel,I come to a differentconclusionaboutthe exogeneityof wageinfla-
tion. The type of model that involvesa fixed planningperiod, based on
institutionalrigidities,modifies the rational-expectations results. Essen-
tially, the institutionalrigiditiesallow the systemto appearto be "irra-
tional"for at least the length of the planningperiod.Hence, the Sargent
statisticalexogeneityconditionsmay hold, but they are relevantonly over
the longertimehorizon.For quarterlyor evenyearlymodels,the empirical
relationshipsmay still be moresystematicthana moving-average, autocor-
relatedstructure.Consequently,even if one adoptsa rational-expectations
framework,plenty of room remainsfor aggregateunemployment-espe-
ciallyif it can be properlymeasured-to influencewageinflation.
To summarize,the Okunpaperis an importantstatementof the inflation
mechanismand its resultingwelfarecosts. The majorfactoris that institu-
tional arrangementsthat are currentlybased on zero, or close to zero,
inflationhave been strainedand are being forced to change. In today's
economy, even if inflationis correctlyanticipated,there remain institutional
and contractualrigiditiesthat preventeconomicactors fromadjustingto
inflationand consequentlythatimplythatcontinuinginflationwill impose
real effectson the economy.
GeneralDiscussion
WilliamNordhausbeganthe critiqueof "customermarkets"by agreeing

withthe authorthatmanyfirmsandindustriesadjustto changesin demand
primarilyby varyingquantitiesratherthan prices.Thatfactneedsa theo-
ArthurM. Okun 399
reticalexplanation;and while Nordhausfelt that he had not seen a con-
vincingone elsewhere,he did not findOkun'spersuasiveeither.Basically,
theamendmentof the paperto the Phelps-Winter modelmakesthe propor-
tion of customerswho go shoppingchangediscretelyin responseto evenan
infinitesimalpriceincreaseby the supplier.To Nordhaus,that was not an
adequaterationalefor the stickinessof pricesin responseto shifts of de-
mand.The maindifficultywasthathe wasnot persuadedthatthe informa-
tion ticketshad sufficientvalueto consumersto makethema quantitatively
significantphenomenon.RichardCooperpursuedthe issueof the value of
informationtickets.On the one hand,he wasimpressedby the information
costs involvedin obtainingservicesfrom new supplierswhen one moved
cross-country,forexample.Thatindicatedthatthe bilateralrentssharedby
cooperatingsuppliersandcustomerscouldbe quantitatively important.On
the otherhand, if those searchcosts are high and the quasi-rentssubstan-
tial, Cooper thought that one should expect well-organizedmarketsfor
consumerinformationthat would attempt,in effect,to supplyOkun'sin-
formationticketsat a lowercost.
WilliamPoole felt thatthe customer-supplier relationshiphad to rest on
the confidenceof the buyerin obtainingthe lowestrelativepriceof a prod-
uct ratherthan the most stableabsoluteprice.If the buyerhas confidence
that some discounthouse consistentlychargesfavorableprices, he will
acceptrisesin theirpricetags, still feelingthat the pricesare lowerthan at
competingretailers.
Robert Hall arguedthat the cyclical constancyof markupscould be
reconciledwiththe classicaltheoryof pricedetermination as long as supply
is highlyelastic,averagecosts are thus flat, and capitalis a mobilefactor
thatearnsa rentalprice.In responseto a questionfromOkun,Hall agreed
that the outputof any particularfirmwouldbe indeterminateunderthose
circumstances.On a relatedpoint, Robert J. Gordon remarkedthat the
absence of market-clearingarrangementsdid not necessarilystem from
informationcosts or customerrelationships,but rathermight reflectlow
short-runpriceelasticitiesof demand.In effect,the hotelownermaynot act
as an auctioneersimplybecausehe doubtsthat he could fill the rooms on
somedaysevenat a verylow price.The classicaltheoryof deriveddemand
could explainsuch short-runinelasticityfor many kinds of productsand
services,without invokingOkun'sparticularrationale.
The strengthand potentialitiesof the customer-productmodel were
stressedby some of the discussants.CharlesHolt felt that this conceptual
approachwouldstimulateotherresearchers to do additionalfruitfulwork.
He suggestedthateconomistsmightlearnmoreabouttheseissuesfroman
extensivebody of literaturein the marketingfieldthatdiscussedtechniques
of developingcustomerrelationshipsand establishingbrandloyalty.Holt
saw importantalternativesfor publicpolicy in attemptingto improvethe
structureof marketsor to intervenein the existingstructureso as to make
it functionbetter. He urged thoroughexplorationof the possibilitiesof
usingtaxesand subsidiesto discourageexcessivewageand priceincreases,
to encouragethe expansionof production,and otherwiseto elicit micro-
economicbehaviorwith favorablemacroeconomiceffects.R. A. Gordon
approvedof the introductionof the conceptof fairplay and similarlong-
runinterpersonalconsiderationsinto the analysisof pricing,as recognition
of an importantkindof rationalitythat was generallyignoredin neoclassi-
cal general-equilibrium theory.He also expressedhis supportfor the kinds
of novel policy approachesthat Okun had espoused.In Walter Salant's
judgment,the paper providedthe best explanationto date of some im-
portantfacts about price-quantityinteractionsthat have been empirically
well establishedbut have often been analyticallyignoredor rejectedbe-
causeno theoreticalframeworkwas offeredto explainthem.
In contrastto his reservationsaboutthe customer-product market,R. J.
Gordon felt that the analysisof careerlabor marketswas on the "right
track."He suggesteda numberof otheravenuesthat might be explored,
likethe inherentlogic of seniorityrulesandthe waytheytendto encourage
layoffsin preferenceto wagereductionsin a recession.Gordonwasencour-
aged by a growingbody of literaturedevelopingthe concept of implicit
contractsin labormarkets.Hall, on the otherhand, doubtedthat specific
humancapital was of sufficientimportanceto explain very much about
wagerigidity.He agreedthatthe existenceof specifichumancapitalcreated
a zone of indeterminacy in wages;hence,wageratesmightnot respondto
changesin labor-markettightness,as long as they remainedwithin the
alteredzone. However,he partedcompanywith Okunon the empiricalsig-
nificanceof specificcapitaland hence on the size of the zone of indeter-
minacy.Hall pointedto the extremelyhigh interfirmmobilityof certain
types of craft workerslike lithographers;they must have essentiallyno
firm-specificcapital.He also suggestedthat, if problemsof internalwage
structure deterred existingfirms from taking advantageof a very weak
labormarketby reducingwages, one would expectthe acceleratedestab-
lishmentof new firms.
A numberof participantstook issue with Okun'srejectionof steady
ArthurM. Okun 401
inflationas a targetfor publicpolicy. RobertSolow contendedthat, even
in a worldof customermarkets,the majorcosts of inflationlay in the irreg-
ularityand unevennessof the rateof priceincrease.He saw no compelling
reasonwhythe relativeirregularity arounda positiveinflationtrendshould
be greaterthan that arounda horizontaltrend. Moreover,the customer
needonly believethat his preferredsellerquotespricesin the rightgeneral
range,not that they be constant.In short, while smooth inflationis ad-
mittedlyimpossible,so is smoothpricestability.R. J. Gordonnotedthat a
cost-benefitcalculationwas necessaryto establishwhether,once the infla-
tion ratereached6 percent,it wasbetterto tryto beatthe ratebackdownto
zero or to try to maintainthe prevailingrateas smoothlyas possiblethere-
after. WilliamFeltnersupportedthe verdictof the paper;in his view, to
accept an inflationtrendjust becauseit had in fact developedentaileda
posturefor publicpolicythatlackedcredibility.JamesPierce,on the other
hand,arguedthat a targetof zero or near-zeroinflationhad a tremendous
dispersionin one directionand thereforemightcreatemoreuncertaintyfor
the publicthana targetfromwhichdeviationswerelikelyto be moresym-
metrical.Pierce interpretedrecent developmentsin financialmarkets-
suchas floating-ratenotesandmoney-market funds-as evidenceof greater
adaptabilityto inflationthan Okunhad suggested.
MichaelLovellcommentedthat the "refixingof arrangements" that in-
flationnecessitated,accordingto HicksandOkun,wasnot necessarilybad.
By shakingup conventionsthatcontributeto pricerigidity,inflationmight
contributeto economicefficiencyby facilitatingthe realignmentof relative
prices.
Okunrespondedto a numberof the comments.In responseto Nordhaus,
he defendedthe discontinuityin the shoppingfunction,arguingthat some
changein priceis qualitativelydifferentfromno changein price,particu-
larlyinsofaras it could affectthe confidenceof the customerthat today's
price would be maintainedtomorrow.Agreeingwith Fellneron the need
for credibleprice-levelobjectives,he contendedthat price expectations
mustdependon the government'spoliciestowardgrainexportsandenergy
as well as on monetaryand fiscalpolicy. He reiteratedthat conventional
wageand pricecontrolsstood low in his rankingof potentialsolutions;he
sharedHolt's sympathyfor techniquesthat mightuse the pricesystemto
deal with macroeconomicexternalities.More generally,Okun urged the
need for professionalbrainstormingto develop socially and politically
acceptablemechanismsthat couldhelpto dislodgeinflationwhilereducing
the requiredsacrificesof outputand employment.

Inflation: Mechanics Welfare: Its and Costs

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Inflation: Mechanics Welfare: Its and Costs

Uploaded by

Copyright:

Available Formats

ARTHUR M.

Inflation: Its Mechanics

ECONOMISTSCONTINUEto be baffled by the mystery of inflationary re-

The suggestedinterpretationdevelopedin this paper builds on past

VintageKeynesianismof the WorldWarII periodoffereda comparative-

dynamicview of the approachto that equilibrium.'Overa wide rangeof

The Phillips approachdepartedfundamentallyfrom vintage Keynes-

rationale rested on a model of "stochastic macro-equilibrium" in which

8. These gave no reason to suspect that the structurehad changedthrough 1968. In

Accelerationismwasthe most fundamental transformation of the Phillips

More generally, the unemployment-inflation experience of the first half

13. See Robert J. Gordon, "The WelfareCost of Higher Unemployment,"BPEA,

Keynesiantheoryof involuntaryunemployment.'7 And the commontheme

PURPOSE OF THIS PAPER

I shallnow sketcha nonclearingargumentthatmakessenseto me, draw-

19. "UpwardMobilityin a High-PressureEconomy,"BPEA, 1:1973,pp. 235-52; and

standingexceptionis JohnHicks,whoserecentanalysisof the welfarecosts

22. Nor is it conceivablethat, despitethe shift in demand,each hotel ownerbelieves

creasediscretelythe proportionof his customerswho will buy information

THE BUYER-SELLER RELATIONSHIP

An establishedcustomer-supplier relationshipintroducesa bilateralmo-

24. For a discussionfocusingon the labormarketthat considersa varietyof informal

26. PhillipCagan,TheHydra-HeadedMonster:TheProblemof Inflationin the United

THE SCOPE OF CUSTOMER RELATIONS

By dampeningits cyclicalpricefluctuations,the customer-strategy firm

Career Labor Markets

Customer-supplier relationshipsin productmarketsare paralleledby

experiencedworkersregularpay increasesand growingseniorityprivileges

Careerjob arrangements desensitizethe wagerateto changesin overall

cost of payingovertimepremiums.But the firmhas a strongincentiveto

CRITERIA IN WAGE DETERMINATION

Variousgroupsof employersandworkersmayopt fordifferentstandards

INFLATIONARY BIAS IN WAGES

In summary, career labor markets make wages less sensitive to labor-

Inflationin the Mixed Customer-AuctionEconomy

A SIMPLE ALGEBRAIC MODEL

A simple algebraic model can describe some of the mechanisms of infla-

< * _ W~~~,> 5 c0.

(2) PCQC (1 - a)7

the inflationin the auctionsectorand the expansionof outputin the cus-

wherek is the weightof auctionitemsin the cost of living.It is convenient

40. These models closely resembleone constructedby James Duesenberryway back

ASSETS AND INTEREST RATES

Wheninventoriesof auctionitems are broughtinto the analysis,they

ADAPTATION AND ACCELERATION

The institutions of the customer sector are geared to a basically noninfla-

Implicationsfor Welfare and Policy

Customermarkets are valuable institutions to society, because informa-

trading in spot and futures transactions and to the negotiation of formal

imposedby intenseinflationon these relationshipsis not enormous,but

Someproposedpolicystrategieswithrespectto inflationseekan alterna-

45. To do so, they would have to outweigh any negativeimpact of deflationaryex-

Another proposedstrategyfor policy would substitutefor the dollar

promptlyandmorefullyto changesin the pricesof auctionproductsandin

49. "AlternativeResponses of Policy," pp. 201-02.

POLICIES FOR THE AUCTION SECTOR

that essentiallyignoredtheirmacroeconomicinflationaryand recessionary

51. See, for example,WilliamNordhausand John Shoven, "Inflation1973:The Year

Themakersof stabilizationpolicyfacea particularproblembecausecus-

Finally,a casefor price-wageinterventionby the governmentemergesin

ternalitiesmay be large enough to compensatepartiesthat accept some

The worldof mixedcustomerand auctionmarketsposes seriousdilem-